Bob, I performed some more thorough testing on my Alleghany turntables since we talked about this, but my findings don't and won't collaborate or support your findings due to a major difference in our implementation. However, I think the results are worth sharing here while we're on the topic. On the COA the TT's at Alleghany and Ronceverte are used only for turning... I only utilized two exits, one to enter and a short run-out track out the backside... there's no roundhouse or other yard exits. I tried to be systematic in my testing, pulling forward onto the TT, turning and pull off, turning and backing off, backing on, pulling off, turning twice, not turning at all, etc, etc and every combination of those actions. I tested with an ES4400, SD40-2, SD70ACe, Challenger, and Britkits Heavy Pacific.
The only failure I could create was with the Challenger... it doesn't fit (or I couldn't fit it), which is where I learned that the [invisible] collision point of the locomotive defines its length on the TT, not the wheel position. Other than that, all of my turns and drive ons/offs worked fine. Btw, to address one possible cause of trouble brought up earlier, I should point out that the TT in Ronceverte has a very short curving approach track, which in fact did take many hours of trial and error to get installed this past Spring. The Alleghany TT is on straight level track and installed without a hitch way-back-when on RW version 114. Things I learned from that were: 1) to be sure not to mix trackrules and/or track types near the TT; 2) set your TT in place then off to the side somewhere draw out a short length of straight, level track, then use the Join Tool to connect it to an exit on the TT. Then cut to length as needed.
The two oddities I observed were that the TT starts out aligned initially, but after turning and then doing it's little drift/pop into final position, it was off-center by half the track width. But that was only visually... driving on/off in either direction and orientation was normal and the wheels followed the path where the rails
should have been. Upon a second turn the TT aligned perfectly, just like it started. The other anamoly I observed was an occasional "hiccup" where the TT would start to turn, but after only 10-15 degrees it would stop and return to the start position. The next command to turn worked normally.
Bob, this has all led me to one thought that I didn't see mentioned in your post...

... what is the spacing of your 14 tracks? Are they an even number of exits apart or random? Is it possible that even though it's a 32 exit model, we might only be able to effectively utilize a certain number/pattern of exits? I wonder what it would do if you only used every 2nd, 3rd, or 4th exit, keeping them all directly opposite each other?