I may be way off the mark here, and it doesn't help that i don't have the wiring diagram in front of me, but i have an idea of what it might be. I also apologize in advance if what follows is a real long explanation of something that's common knowledge to everyone else...
At the risk of sounding too obvious, the regulator... er, regulates the voltage going through the system by comparing the input voltage to some internal benchmark voltage. (Note that if you're still using the zener diode, that acts as your regulator) If it determines the voltage coming through from the alternator is a "healthy" voltage it allows it through to charge the battery, as well as going to the assimilator which then sees appropriate voltage and, therefore, a properly-operating alternator.
However, if the voltage is too high it does either one of two things, depending on the regulator type. It can either short SOME of the current to ground to bring the voltage down to an acceptable level that won't toast the battery and various electrical bits living on the system, or it'll just act like a regular switch and short ALL current to ground until the input voltage drops back down to a reasonable level.
With your steroided hi-output charging system, i'm guessing that at low revs and when idling the system charges fine and the warning light stays off, but your revs don't have to go up too high before that high voltage threshold on the regulator is met and it starts shorting everything to ground. I'm also guessing the series goes alternator to reg/rec to assimilator to warning light, and in that case the assimilator won't see any voltage coming from the alternator since the regulator's shorting it all to ground. Then the light goes on because the assimilator thinks the alternator's dead, when it's actually just working TOO well and the regulator is stepping in to protect the battery from getting fried.
Of course, this is all assuming the regulator on your bike is one that dumps everything to ground once the upper voltage limit is met, or dumps enough that the voltage going to the assimilator is low enough to set off the warning light. It's also assuming the assimilator is just connected to the regulator and not tied into the alternator somehow and directly reading its output. I think both of these depend on whether you have a newer reg/rec setup like Podtronics or the old Lucas rectifier and zener setup because they're wired up slightly different. This is my basic understanding of this stuff applied to just the Commando, so somebody let me know if Commando/Lucas components act differently than this. They ARE Lucas components after all...
Not that this helps fix the problem or anything either, but if i'm right at least you'd know when the light comes on at higher revs that it's not because the alternator's going bad...