games such that
they reach an opposite-colored KBPP vs KBP endgame
and
there was at least one result-changing error after that point
and
I posted the tablebase report of what those errors were
Honorable mention goes to A Pena Riasco vs J Henrichsen, 2013, in which, on 3 consecutive moves, black could've drawn by going into an opposite-colored KBPP vs KBP endgame, but when black finally did so, it was lost and white kept the win from that point on.
The "usual situation" for this collection is
connected passers vs. passer, with at least 2 pawnless files in-between.