The tests here were easy to adjust (partly because I know that tests tend to be deletable in this repo and shouldn't be given too much respect) but in other places they place significant obstacles to doing PRs.

Someone making a PR encouters a failed test, and then it's not clear whether the test indicates a problem in the PR's logic. And it's not clear whether it's his responsibility to get the tests to pass by changing the logic, or his responsibility to get the tests to pass by changing the tests, or @Happypig375 's responsibility to get the tests to pass by changing the tests.