Picture a rotating Harley crank assembly. Now picture the left (drive) side being snatched to a halt (a worst-case scenario)....What will the right side want to do? It's pretty heavy and it's only pumping oil and operating the valves through the camshafts. Won't its momentum tend to make it want to continue rotating?....
It's easier for me to picture this happening by hard deceleration of the left wheel than by merely operating at a lower engine speed on a regular basis. Hard deceleration of the left wheel is easy to see happening by sudden downshifting without matching engine speed to the new gear, or by somehow locking up the rear wheel. It's not so easy for me to see it happening sufficiently because of motoring down the highway at 2k RPM and adding some throttle.
I believe that the intensity of the snatch force (torsional vibration) increases dramatically as the engine is lugged down and the compensating sprocket begins to hammer against its stops. My sense is that 2,000 rpm is probably outside of the critical range, but 1,300 to 1,500 rpm in 6th gear would probably get you there with an aggressive throttle setting.
The shift occurs when the right flywheel's inertia overcomes the interference fit of the crankpin. That is to say that in this scenario the rotational energy stored by right flywheel overcomes the friction at the crankpin, allowing the right flywheel to continue rotating while the left flywheel rapidly decelerates; essentially what glens said in his post.
While it is certainly possible to shift a set of wheels with behavior like a clutchless upshift, (rapid downshifting would increase the crank’s speed) it is my understanding that lugging an early Cruise Drive, as described, will produce load spikes of 1,400 - 1,600 foot-pounds at the crankpin in a stock Twin Cam, which exceeds the capacity of that joint. JMHO