intel/skqp rare one pixel failures
We recently noticed, that very rarely skqp tests fails with one different pixel in random location. I cannot reproduce it locally.
Usually it's one black pixel in random location in the rendering, also test seems to be random.
Examples:
- 1st run (job): https://gitlab.freedesktop.org/mesa/mesa/-/jobs/23083107
- 1st run (report): https://mesa.pages.freedesktop.org/-/mesa/-/jobs/23083107/artifacts///results/gl/report.html
- 2nd run (job): https://gitlab.freedesktop.org/mesa/mesa/-/jobs/22904999
- 2nd run (report): https://mesa.pages.freedesktop.org/-/mesa/-/jobs/22904999/artifacts///results/gles/report.html
/cc @anholt