I suspect this is the point where the engineer misses the issue that the physiology means that the energy needed to do the work of loading the potential energy is greater than that potential energy due to the inherent inefficiencies of muscle.

I'm neither a physiologist nor an engineer, but I can figure out that if there is energy loss of any kind per revolution, this energy loss is going to be greater at 90 rpm than 60 rpm. My advantage is in being able to put out a lot more torque per revolution on a sustained basis than "normal", so for me it would be unnecessary to reduce the gear and increase the cadence with all of the consequent energy loss per revolution.

BTW, I do vary my cadence a lot on a ride, I often spin a relatively low gear on hills, and will spin a low gear for a short burst to accelerate initially from a stop. I also will ride with a faster cadence in a lower gear into a headwind, and occasionally will do something similar in great heat. Not sure why, but I suspect the faster moving legs are a bit better at dissipating heat. I don't know if the higher torque actually generates more heat internally, but that could also be the case. These could easily be additive factors.