"Nibor wind" sums the situation up perfectly and there is no big issue in designing a controller to track the input power cubed, just that I don't know of one available commercially.

The simplest heater control is phase control with thyristors or triacs but the normal firing circuits work at fixed frequency, I haven't seen anyone do it with wild variable frequency.( I have some ideas but have never actually tested the idea).

The easiest solution is to rectify to dc and use a mosfet or igbt as a buck converter to load the heaters. You could measure prop speed ( frequency) and use a multiplier to get speed squared and track power to this but I still live in the dark ages of analogue control ( and it works). I am sure the digital boys can adapt the idea to digital control but this is probably way too complicated for you.

I think you could adapt Hugh's circuit to do what you want and it should do it well enough with about 3 steps of load increase. You won't get good results with a single fixed load.

The capacitor idea is fine in theory but unless you design a special high frequency alternator the capacitors get out of hand. Motor run capacitors aren't the world's most reliable things and electrolytics ( ac or back to back dc) are something that I wouldn't trust long term at any power level.

In theory heating control is simpler than battery charging but in reality battery charging is easy as long as you don't want good performance. The battery constant voltage reference keeps the mill under control. A fixed resistive heating load doesn't.

http://www.folkecenter.net/mediafiles/folkecenter/pdf/Small_Scale_Renewable_Energy_Control_Systems.p df