OK, the first issue is time lag... especially in a floor. There is no need to set and reset every second, every minute will do. Estimating the needs of the house is relatively easy with a well defined program including hysteresis. The problem used to be with single speed pumps (solved) and single stage boilers (solved...to a degree with condensing and modulation, negative pressure gas valves and all that) and with inverter heat pumps (solved to a lesser degree due to the tighter operating tolerances of HPs in general over boilers).

What happens when a really cold wind comes up sucks the heat out of a house within 10 min? No floor heat will compensate easily for that but it certainly does help. I did a loading dock and warehouse for a national supplier and as soon as the door opened, it froze in the building but as soon as the door closed 30 sec and it was warm, but they had 30cm of concrete to hold the heat. That is the same idea as the buffer tank in a way.

As much as I would like to reduce flow rates in the floor to save on pumping power (and tubing degradation over time) I see your point about higher flow rates. Buffer for defrost, low floor temps to keep COP high and a high enough flow rate to ensure that we don't get a 12C delta T over the floor. The slower you go the more liquid temp is lost over the circuit length.

That said, it is difficult for any system to accurately predict the weather future. This is great fun, far more interesting than boiler work.