imported/w3c/web-platform-tests/css/css-transforms/animation/transform-interpolation-rotate-slerp.html There is a slight difference in some edges (including the edge between green/blue portions) when the test case interpolates with the inverse. The deviation seems to be affected by the position, varying slightly according to the position of the rectangle (i.e. replacing all 5 rectangles with the same transformation yields different mismatches).
Update: It's actually flaky failing. Running locally, 6 out of 100 runs passed. But it's still related to inverse interpolations. Without them, 100/100 passes. PS: Updating the revision that introduced the regression. Mixed it with another gardening patch.
The test or tests filed under this bug are not failing anymore. Test expectations updated in https://commits.webkit.org/258724@main.
<rdar://problem/104077189>