Based on comment 0 (which described a difference between the testcase/reference rendering), I think that was an unexpected test-failure where the image was failing to draw. Whereas the recent instances of this seem to be unexpected passes, where the testcase and reference case are rendering more-identically-than-expected. These recent instances are likely to be fallout from a recent infra change that had somewhat-unexplained theming-related side effects. See bug 1894445 for another similar case. (In this case here, the fallout is benign-ish and we could just adjust the fuzzy annotation, potentially.)
Bug 1812425 Comment 8 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
Based on comment 0 (which described a difference between the testcase/reference rendering), I think that was an unexpected test-failure where the image was failing to draw. That failure probably did go away over time (when this was marked as INCOMPLETE 6 months ago). In contrast, the recent instances of this seem to all be unexpected **passes**, where the testcase and reference case are rendering more-identically-than-expected (vs. the fuzzy annotation in [reftest.list here](https://searchfox.org/mozilla-central/rev/e69f323af80c357d287fb6314745e75c62eab92a/image/test/reftest/downscaling/reftest.list#177), which has `fuzzy-if(gtkWidget,18-19,5502-5568)`). These recent instances are likely to be fallout from a recent infra change that had somewhat-unexplained theming-related side effects. See bug 1894445 for another similar case. (In this case here, the fallout is benign-ish and we could just adjust the fuzzy annotation, potentially.)