It turned out that in the rush to finish Candy in time for the Motorama contest in February, I messed up the motor winding order and had a few shorts in the windings (this was done a day before competition). I probably also messed up the heat treatment of the steel teeth since I had to cut it short to make it to comp on time. When I had some time later, I took apart both stators, added another insulating coating to each layer, and re-wound them with more care. Then I joined them in the proper order (dLRK scheme).
The concept works! The outer ring spins up flawlessly by simply connecting to an off-the-shelf sensorless controller. As much as I would like to enter Candy in the competition, there is still the issue of the batteries not fitting properly (part of the reason I had to hold down the top during the spin test), no power switch, and wires too close to the inside of the spinning ring where they could get cut. Maybe if I have some time later I will fix that stuff, but now that the concept of a direct-driven ring spinner has been proven I am starting to see a bunch of flaws in the design and am less excited about it.
Check out the spin test video. I bring up the throttle slowly as I am not sure what sort of power to expect from the motor. Once it turns on it is actually quite impressive, I can feel the torque on the robot from the accelerating ring. The maximum I had the courage to test while holding it like this was about 20% throttle.