The tests in tests/testing_tools/adapter/test_functional.py compare the output of running pytest test discovery against a (rather long) hardcoded list of nodes. Not only is difficult to debug, but it also makes the tests brittle (see when the Windows behaviour changed and we ended with a CI failure: #12372). We should refactor these tests in a way that is not dependent on pytest's output.

💡 We should probably refactor our Python unit tests in general, and not just these ones.