You are right. I just conducted the experiment. I made a sloppy 90volt pack from scrapped cells to test it. It only regens up to 84 volt and above it it releases the motor. I dont really know what to do tho. I ordered 714 cells for a 21S34P config. I hope I can cancel 34cells. Tooo baad that extra 200 rpm would have come useful. It would have been "free" power.
The bright side tho is that I might not have to squeeze the pack so tightly this way...I might also be able to use a honeycomb holder this way so I dont have to glue the cells. However I want to make a water cooled pack and the gap of the honeycomb holder is too large it uses up too much water adding extra superfluous weight, so I might just stick to the gluing with a slight added heat shrink spacer between the cells. Also with the honeycomb I should also give up 2Ps too cause of space issues.
Anyway thanks for the great info had you not told it I might not have done the experiment...