If I had been paying attention I would have know that the existing
control does feedforward. Thanks again Bill.
I disappeared down a large rat-hole investigating this which I don't
have time for. I think I may have learned a few things on the way.
1. Modelling engine power from throttle position is too difficult to
be reliable.
2. Measuring engine power is very possible for electric motors. This
allows control of engine power instead of throttle.
3. Using estimated propeller thrust efficiency (Kg/W), motor
efficiency and ESC efficiency, you can get a thrust control loop
instead of throttle control loop.
For those already measuring battery current and voltage, this method
would come for free.
All this is going to have to wait (again) until I have more time for
it.