Coming back to measuring the disunity, it becomes apparent that simply breaking down the net KE into its component KE's reveals further context in which to interpret the momentum deltas:
For instance note that the wheel's KE isn't increasing even though the radius of the blue disc is increasing; implying that the wheel motor sees a constant MoI, as if the blue disc were remaining at fixed radius, so long as the green one it's mounted to is rotating.
This is thus one area where my momentum calcs were derailing, since i was including the blue disc's point-mass mr² in the grey wheel's net MoI calc. I was also however including its mr² as relative to and part of the green disc's MoI..
..i could see however that the net KE components only count the blue trans KE once of course, hence trying to calculate two different dL's for it was obviously a mis-step.
But besides clearing that up, it's also revealing to see the blue and green dKE's separately, since blue rotational is a load on the k-motor, whereas its translational dKE is being powered by CF-PE as it's pulled outwards, hence a load on the w-motor.
Hence this sharpens up our view of the gain mechanism - of what the exploit is, and where it's arising.
Here's the max-freq results from the above Q1:
kTa = 12.8330934
wTa = 4.704379623
CFPE = -0.999937481
net in = 18.537410504
weight dKE = 23.50718262
rotor dKE = 1.79549284
net dKE = 25.30267546
diff = +6.765264956
CoP = 1.36
For context, here's the contents of the KE meter:
Weight = 0.5*body[21].moment*sqr(body[21].v.r)
Rotor = (0.5*body[5].moment*sqr(body[5].v.r))+(0.5*body[21].mass*sqr(body[21].v))
Wheel = (0.5*body[1].moment*sqr(body[1].v.r))+(0.5*body[5].mass*sqr(body[5].v))
..these summing to meet
kinetic().
So the most-striking detail we notice is that the blue rotational's 183% of the work done by the k-motor - the spin-up evidently having been subsidised by something.
The rotor dKE is obviously the blue translational KE as it's reluctantly pulled outwards by CF force, against its efforts counter-torquing back in the opposite direction.
The output CF-PE is 1 J, the rotor dKE is 1.79 J, and the total load on the w-motor is 4.7 J. Subtracting the former from latter leaves 2.9 J of work done by the w-motor that must be contributing to the blue rotational dKE.
Again subtracting the former from latter leaves 20.59 J of dKE attributable only to the kTa of 12.83 J, so an excess of +7.76 J and a k-motor CoP of 160%.
Looking at the velocities, the rotor actually decelerated slightly over the drop, from its initial 2 rad/s at TDC, down to 1.85 rad/s at 9 o' clock, hence the KE gain isn't attributable to a reactionless acceleration of the green rot / blue trans component.
That only leaves the inter-reaction - or lack thereof - between the two motors; the system's OU because the k-motor isn't applying a reciprocal compliment of counter-torque upon the w-motor when spinning up the weight.
The wheel motor's basically somewhat agnostic about the rotational acceleration of the weight..
..seemingly consistent with the hypothesis that counter-torque from the spin-up is effectively being sunk to CF force and time..
