A personal shoutout from a dev; nifty!
Anyway, a couple of days after I posted that thread I did manually compute the diagonal FOV from the horizontal and vertical estimates from my program. (No need to change the program.) I can confirm that I got 74 point something which is reasonably close given the approximations involved. I’d never heard of anyone actually measuring FOV diagonally, though, so I figured a simple bug was the more likely explanation. The move to Hor+ style scaling is welcome news.