I have been too busy to follow this closely.
If you want heating only then why do you want a battery. If it is partly for battery use with normal loads and the rest used for heating it may make sense.
The best way for heating would be 3 phase balanced load heaters but that needs a lot of heaters or a complicated controller.
You can do it on the dc side but you are left with dc contactors or mosfets or IGBTs as the switches.
I can't imagine you will need a lot of stages to load the thing near enough, the prop has a great deal of latitude.
You need the dc voltage to rise roughly with wind speed, clamping to the constant voltage of a battery is a disastrous way to go at things except for pure simplicity.
Stage one lets it freewheel up to a point here the blades come out of stall. You then start loading and a suitable resistor will probably keep you near enough on track from cut in to perhaps 15mph. If you choose this resistor right you will not have enough load for higher winds so you need to switch in more load. Best way would be to switch in one more resistor that would give correct load at furling point and pwm control this so that it was not too steep in the mid wind range.
If you want to avoid pwm then most likely you would need 2 more stages after the first resistor, something that in parallel with the first R would load you properly perhaps to 20mph. The final stage would add one more R so that the 3 in parallel would give the final full load loading.
All these stages will need hysteresis and whatever you do they will cycle on and off and you will have a crude low frequency pwm. If you use contactors the hysteresis will need to be large to stop them cycling too often and hammering themselves to death.
For mosfets the things can be left to cycle at relatively high speeds ( once or twice a second at the worst points would be ok. You should get stable periods where no frequent cycling takes place but with the nature of wind I don't think it will stay in any one stage for long.
Probably the best thing to reference is prop speed ( frequency) and a 2917 tacho chip would give a voltage proportional to prop speed and that could then operate the 3 level detectors to switch the mosfets. The individual levels should be adjustable and you may need individual hysteresis control but it wouldn't need to be much more than selecting resistors in the initial stages of experiment.
It may work using dc voltage for the sensing but the alternator internal resistance will make it much more difficult than speed.
You will need to drive the mosfets sensibly but you will not need the sophistication needed for high frequency switching but it would probably still make sense to use mosfet drivers rather than high impedance things. The resistors and associated wiring will have inductance so a freewheel diode across each would be a wise precaution.
Flux