When working on other things, I found that our coverage tests were failing on master on macOS. I quickly fixed this in #7972, but there's nothing stopping us from hitting this problem again in the future. I think we should avoid this because it can be a confusing/frustrating experience, especially for new Certbot developers, to investigate test failures on their local changes only to find the tests don't even work on master.

To resolve it, I think we should run our coverage tests on macOS in CI. I think this can be done instead of our normal unit tests since our coverage tests just run our unit tests while measuring code coverage.