I figured I'd try to trim down the tests in my pip PR a bit, and I managed to combine two pairs into two parametrized tests. I don't have a good sense of whether it makes sense to collapse things more aggressively. I think if I went much further, I'd end up with tests that have completely opposite assertions depending on a flag, which is a little scary to me. A little variation is... fine... but I don't know about "If this test flag is set, the installation should succeed, otherwise it should fail".
I guess I don't have anything else to say here. I think there's more work to do, but it's not happening now.