Looking at the older R32 models, even there you see differences in COPs and SCOPs, with newer models squeezing out better performance. Seems likely the same will happen with the R290 models, better COPs and SCOPs with newer models.

The other aspect is modulation, and getting better performance by being able to modulate down. Based on what you read and see from performance monitoring, this is still a problem on many of the larger capacity heat-pumps, good modulation range is not yet universal, so that's also likely to steadily improve over time. Again, helping raise performance.

Then there's all the other functions which will likely get better - the system for defrosting is pretty primitive, that will get improved. With better modulation and improved self-defrosting, I reckon the need for a buffer tank will likely disappear and using existing cylinders will start to become viable with higher temperatures and better modulation.

At some point heat pumps will become essentially a drop-in replacement for an existing boiler, taking away a lot of installation difficulties and cost. 

Much of that is true in principle, but its not likely to be revolutionary. 

You are also correct about the fact that heat pumps don't reach theoretical efficiency, in fact they are almost all within a few percent of 50% of the max possible according to Carnot's law.  However the key components are an electric motor and piston, which have both been around for a century or so, and thus we are unlikely to see revolutionary changes in these either.  Diesel engines have been around since 1897, but are still only 30% efficient! 

Bear in mind that heat pumps have been around for nearly 100 years and residential air-conditioning (which is based on heat pump technology) well over half a century ago.

So I cant see that your argument holds for heat pumps any more than they hold up for any other modern appliance.  Sure there will be continuous improvement, particularly in the control systems (which is the weak point of many current heat pumps) but nothing that will make a revolutionary change within the lifetime of a typical boiler.  What it will do however is make it more fool proof for installers, by automating the adjustment of (eg) weather compensation and radiator balancing.  Whether it will stop installers adding in unnecessary external controls, buffers and other performance crippling 'features' remains to be seen, hopefully so!

If all your purchase decisions were evaluated on the basis that there will be evolutionary improvement, you would never buy a car, mobile phone, computer, or television (and many other modern things).  I would be willing to bet you have at least 3 of these!  If your car fails today, how long are you willing to wait for a replacement? 

So faced with a choice of replacing a failed boiler (which is OPs situation) I don't think you can realistically argue 'future technological advances' as a reason not to select a heat pump (and thus wait say 15 years) provided you are prepared to ask the right questions, as OP clearly is.  There are several other arguments, but not this one!

Will heat pumps ever be a drop in replacement for a boiler?  Its an interesting question.  New building regs mandate a max design temperature of 55C and, for those houses, I would say, yes.  But there is still the majority of legacy housing stock to deal with and, like it or not, a fair few have heating systems that rely on high temperature operation.  That's not sensible for reasons of comfort (low temperature heating is generally much more comfortable) nor for reasons of efficiency, even with boilers. 

Unfortunately our, in part undereducated, heating industry has stuck to fixed (high) temperature heating when the more enlightened European countries moved to low temperature weather compensated systems at or around the time that condensing boilers were introduced.  As a consequence of the backwardness of our heating industry we are paying 10% more for our (gas) heating than we need to and enjoying lower levels of comfort.   Eventually this has to change and that inevitably involves some changes to legacy heating systems. 

I also think that MCS is over prescriptive on some elements of system design and that there definitely opportunities to reduce retrofit cost.  That said, doing so requires our heating industry to become more innovative/intelligent and, whilst there are clearly installers that are able successfully to think 'outside the box', there are equally installers who, even given fairly rigid rules, manage to screw up.

Taking this all together I personally feel we will get quite a lot closer to a drop in replacement, but it will be a sad day if we get to the point of a literal drop in, because we would be missing some tricks that we failed to learn early in the 21st Century.

Overall I do accept that there are arguments for delay, particularly if you are an uninformed purchaser and there is no immediate imperative.  However OP is neither of these, he has a failed boiler which must be replaced and is asking the right questions which, if he can make some time, should result in a heat pump system that delivers greater comfort for about the same running costs at current oil prices (oil prices are very volatile and currently quite low, so its difficult to predict the future) and emits one quarter of the carbon dioxide.

I find the logic in the second part of this difficult to agree with.  Perfectly good high temperature heat pumps are already available, namely many of the R290 ones.  However the basic thermodynamics isn't going to change, so running any heat pump at high temperature is never going to be efficient (nor as comfortable) so 'high temperature' is largely irrelevant.  More to the point that is, very sensibly, not what you intend to do.

Many of the R290 models have only launched in the last year or two, some manufacturers are still in the process of launching their ranges, so these are all first generation technology. Sure they work and the physics remains the same, but it is not hitting the limits of the physics, like all technology it gets steadily better over time. 

Looking at the older R32 models, even there you see differences in COPs and SCOPs, with newer models squeezing out better performance. Seems likely the same will happen with the R290 models, better COPs and SCOPs with newer models.

The other aspect is modulation, and getting better performance by being able to modulate down. Based on what you read and see from performance monitoring, this is still a problem on many of the larger capacity heat-pumps, good modulation range is not yet universal, so that's also likely to steadily improve over time. Again, helping raise performance.

Then there's all the other functions which will likely get better - the system for defrosting is pretty primitive, that will get improved. With better modulation and improved self-defrosting, I reckon the need for a buffer tank will likely disappear and using existing cylinders will start to become viable with higher temperatures and better modulation.

At some point heat pumps will become essentially a drop-in replacement for an existing boiler, taking away a lot of installation difficulties and cost. 

The Mitsi R32s are a little cheaper, do they have the same cascade capabilities?

I think so but I don't know for certain.  A cascade as grant describes it should cover 4-20kW without having to cycle, which again hopefully should be good for efficiency.  I might suggest to swap master and slave once a year (or two) to even out wear!

When using two R32 Aerona³ heat pumps connected in a cascade configuration, one of the two (the supplementary unit) will be controlled by the other (the lead unit).

Under this hydraulic arrangement the lead unit can be used to provide both Heating and DHW, but the supplementary unit will be used only for Heating.

Unfortunately, as per the above, it is not possible to supply two separate flow temps simultaneously. Normal operation would be to target the higher temperature circuit ( Rads ) and use a blending/mixing set to control the UFH circuit.

*Please be aware, Grant UK R290 heat pumps are not currently suitable for cascade configuration.

Fair enough.  As I suggested above there are all sorts of games that could be played,  but which are easily doable depends on the capability of the controller, which may vary widely  As you are considering a cascade this is may be an important area of study unless you want to tinker with home assistant etc. 

I doubt the grant R290 pumps are inherently impossible to cascade, but it's entirely possible that the controller doesn't have the required features.    Cascades are unusual in a domestic context so there is little market for the software.  In the commercial world cascades are relatively common but so are sophisticated building management systems.  In fact the simplest control is probably exactly the same as a supplementary heater, off of the same contact and software, just switching in the secondary when the OAT goes low enough.  I wouldn't be surprised if that's what the grant one is based on their description.

Of course you could have 2 separate systems not cascaded, one for upstairs one for down.  That would allow 2 separate flow temperatures.  However you would lose the modulation depth advantage.  Possibly some clever plumbing could overcome this off course.  Unfortunately my knowledge of cascades in practice is very limited although I do seem to remember that Mitsubishi are a go to, perhaps their cascade controller is particularly good?

When using two R32 Aerona³ heat pumps connected in a cascade configuration, one of the two (the supplementary unit) will be controlled by the other (the lead unit).

Under this hydraulic arrangement the lead unit can be used to provide both Heating and DHW, but the supplementary unit will be used only for Heating.

Unfortunately, as per the above, it is not possible to supply two separate flow temps simultaneously. Normal operation would be to target the higher temperature circuit ( Rads ) and use a blending/mixing set to control the UFH circuit.

*Please be aware, Grant UK R290 heat pumps are not currently suitable for cascade configuration.

Hence why I am always very cautious about surveyed heat loss most particularly in older buildings that have had miscellaneous fabric upgrades and other difficult to determine factors.

PS - Back to cascades - I think Mitsubishi are one of the go to manufacturers.  If you got the 11.2kW and 8kW R290 - the latter of which has 2 compressors 2kW fixed and 6kW variable - you could go all the way from 19.2kW to 2kW without cycling, given the right control.  Now that would be a smart way to run a cascade heat pump and its more or less guaranteed to cover your heat loss both at top and bottom ends.

Yes assuming the pipework permits this.  Not sure this is the most efficient way to run even in the shoulder season though, probably better off just one of them running closer to full chat.  Yet again the answer will depends on what your actual loss is, as well as which make you choose because they play games differently with the compressor efficiency curves.  There again it might just work out most efficient year-round, because most of your heating is at the lower temperature. 

As I say there are plenty of games you could play with two heat pumps, assuming the controller permits it (or you are willing to tinker in Home Assistant or similar).

It matters to us because we're on the tipping point between single 17kW or the dreaded 2 x 10kW, but will we REALLY know until we actually do it and run it? I am tempted to buy a SlimJim for the winter (say the 10kW) and run it at different temps through the winter and see how we fare. As previously mentioned we do have 3 wood burners as well in the mix.

Would this SlimJim idea work (and how would it work with the UFH at 35 degs and rads at (say) 50?)??

Honestly I doubt, based on your oil consumption** you are on the tipping point between 17kW and 2*10kW, much more likely you are on the tipping point between 17kW and 12-14kW.  However only a real measurement will tell for certain.

An electric heater will work of course, correctly plumbed, just bear in mind the running cost.  If you want two different temps you must have a mixer, thats true whatever the heating as there is no boiler or ASHP available that can output 2 flow temps at the same time!.  An electric heater is also a fantastic way (frankly the best way) to determine your heat loss assuming you have a smart meter and can get half hourly meter readings.  You wound need to take into account your log burners of course (you might have to weigh the logs you burn each day for a couple of weeks), but with the right 'experiment' its the most accurate determination available.  Even the best survey is partly speculation!

And would running 2 x 10kW be very much more expensive than a single 17kW (we spend £2500-2800 annually on oil). 

Correctly set up and controlled almost certainly cheaper to run than a single unit, because you will get much better modulation depth.  You could, with the right plumbing, play lots of games, eg arranging them to supply two separate flow temps simultaneously, arranging so that one is a fall back when one fails.  Its actually a jolly flexible system BUT makes plumbing and control more challenging, so needs a sound installer and a make/model that has flexible control for cascades.  That said for the 4K or less that the additional unit costs its a pretty fail safe way to go

-----------------

* one question occurs which might affect this assessment.  Was your quoted oil consumption for years when you were making extensive use of the log burners?