The green square appears to overflow by 25% from expected.
Link to the WPT result:
This was fixed in r273955
This may have slipped by our test suite because these flex-minimum-height-flex-items-XXX.html tests are marked [ImageOnlyFailure] in TestExpectations:
After looking a little further into this, it looks like the test changed to match Chrome's behavior in Oct 2021, 9 months after r273955
The commit message for the WPT test (https://github.com/web-platform-tests/wpt/commit/2c83d2e8d545706b565dc6e3dcf77daa77eb5e96) indicates that the behavior of Chrome, FF, and WebKit was out of spec according to the discussion here:
When running Safari against that revision and the test as it existed in 03/2021 it looks like the revision did cause the test to pass but the interpretation of the spec may have changed. I'll look more closely at the csswg discussion.
Firefox bug about it