Ok, here's the result of my evaluation:
The above graphs show the quality of 50 shots each, with a target at a distance of about 60-70ft and at the shutter speeds shown (green: excellent, light green: good, light red: acceptable, red: poor). The two greens mean keepers, light red you might keep if it was a once-in-a-lifetime shot, red is a definite throwaway. Take the differences with a grain of salt: green vs light green, for example, reflects very small differences and thus is subjective; the differences between a keeper and one of the other categories was much clearer, though, so the graphs allow a good comparison between the different lens setups.
Graphs on the left are lens only, on the right with a TC-14E III.
Interesting effect: the 500 PF seems to have a VR issue when shooting between maybe 1/50 and 1/125: VR works better, in fact, somewhat better than the 200-500's, outside of this range, but results are pretty disappointing within it. I initially thought I made some kind of test mistake here and even re-took a series of shots, but the pattern remained the same with and without TC, so I'm sure it exists.
I guess the OP must have run into this effect and not tested the 500 PF's VR at slower speeds, where it improves again. Bottom line: the VR on the 500 PF is not as consistent as it is on the 200-500, so I'm hoping Nikon will fix this through a firmware upgrade, but its overall performance is pretty promising.
Off to some more testing...