After discussing with someone, I realized that the tests actually test the Python implementation twice, and never the C implementation (since I use traceback.print_exc() to "test" the C implementation).

I don't think it's possible to both catch the output from the C implementation and prevent it from exiting out of the function prematurely, but if anyone knows how to do that, be my guest! In the meantime, I'm going to mess about with _testcapi and see if I can add something there to help with this.