Thank you for bringing this minor outlier to my attention. I appreciate that as a newcomer I have yet to establish any authority on the topics at play here so I'll attempt to give my interpretation of what we're seeing here.
First, I will trust Kodak's rigorous testing of their own film and developer over anything I see from any other company, as the combo of Tri-X and D-76 is decades old; saying it is tried and true would be quite the understatement. From that, what I don't know is how Flic Film's testing was done to obtain that one dev time. Could be precise sensitometry, could be ocular negative appreciation. While I don't have Tri-X on hand to test BW&G with it (and don't plan to acquire any during a trade war), it sounds plausible to me that the dev time on BW&G's datasheet comes from an estimation following a thin negative at a shorter dev time, and/or that from extra caution they lean on the latitude of the film as a fudge factor to make sure people don't underdevelop their Tri-X with their product. Who knows.
Second, as for the equation I provide in the OP, it was an aggregate fit of all the data I could verify as true regarding BW&G and other film's dev times in D-76 1+1. The result is this curve which correlates existing dev times by 97.9% (r²=0.9798) which is excellent, and to me suffices for the equation to be used as a good model of the BW&G dev times. So of course since it's not 100% there will be some variation and some dev times will be a little over or under but that's part of what I call a 'starting point'. The differing trend between Tmax and Tri-X could be the result of that +/- versus the curve, but I don't have much incentive to verify that manually, especially that these films are both well-known and have published times for both developers.
Third, to prove that it works in the real world I went further and tested it with films of various speeds and expiration dates and was most of the time spot-on, first try. Hopefully that demonstration proves that the pudding is in the... hmmm... fixer? (Nope, that fell apart.)
Anyway, I hope this might address most concerns that came up when looking at the data at a more surface level!
Cheers,
-Jason