[BUG] OptiX CI build has broken #2040
Description
For the last several days, all branches (main and every PR in progress) seem to fail the OptiX test job, with lots (all?) of the render tests not matching.
There was never a PR that we merged that was failing this, in fact, it passed CI for the PR #2034, but failed the CI run that kicked off after the merge (which should have been identical) and has been failing ever since.
Furthermore, this morning, I pushed a commit from several days earlier (before any of these failures started happening), to force a CI run -- and it fails, too. In other words, commits that passed a week ago are failing now. Seems to point to something having changed on the GH runners themselves? (Which happens with annoying frequency.) Which also means that it's likely to not be reproducible on our local machines, we're going to have to painfully diagnose and fix the problem by pushing one experiment at a time to the CI runners, most likely.
I'm not sure I can get to this for a couple days, so if anybody wants to take a stab at it...