looks a bit confusing to have fails in CI for all PRs let's create a new pytest tag macos_bug or something, and just skip those tests in CI on mac, will make it easy to see where issues on mac are as well