One shot below is with a 50 and cropped and the other with a 200 and cropped.
If I had to guess, the camera was on a tripod and the one with the 50 is much more cropped than the one with a 200. As such, if you also stopped the 200 down a lot more than the 50 (say, f/2 with the 50 and f/8 with the 200), the only difference between the two images would be resolution (and either motion blur if you compensated with shutter speed or noise if you compensated with ISO). And, scaled down for the Web, neither is going to be apparent.
If I'm right, then a coin toss is going to be as good as anything at determining which is which, so I won't even bother.
More interesting for this discussion would be full-frame, uncropped comparisons "zoomed with the feet" to simulate the actual reason one would choose the one lens over the other.
You have the correct technical answer. They were shot at an identical distance (and the 200 was f/8 and the 50 f/2.2, FX added to mask resolution and other differences - you really know your stuff!).
The point was to illustrate the fact that even routine cropping and framing differences with a particular lens will in fact fairly drastically alter the so called "proper perspective". And if we define what "proper perspective" actually constitutes and then express it in terms of focal length - which has been going on for ten pages now - my point is that it is a complete ruse. Because if, just for example, the so called "proper perspective" is achieved with a 90mm for a tight head shoulder shot, then the facial proportions when increasing subject distance only enough so to shoot a full bust/waist up shot, become equivalent to being shot as a tight head/shoulders with roughly a 180mm.
All of that said, I think you have the right idea of what is useful in actually choosing. And I think the findings would be that a range of lens lengths work well. And some have a look to them that really is not so much about length as other factors. JMO of course.
 forgot to add, those are D cells shot from about six feet. They are about the same height as a typical adult ear and nostrils to brow distance. The angle created roughly the same distance and depth from the left most to right most batteries as with adult nose to ear. All to say, a relevant model for the discussion.