safety: run mutation tests on everything

[adeebshihadeh]

We only run the mutation on the diff against master for speed reasons.

Stats from running on everything and just running ./test_toyota.py -f instead of the full suite. It's definitely fast enough if we can order the test properly, and actually should be faster than what we do now since we don't fail fast.

[info] Mutation score: 78%
[info] Total execution time: 110093ms
[info] Surviving mutants: 42

real    1m52.382s
user    5m57.402s
sys     0m11.549s

---

Turns out mull-runner does a "warmup" pass and filters out code paths it knows won't get hit, so it's not quite as fast as I thought. I'm adding the --include-not-covered flag so the coverage doesn't depend on the test program at all.

We have 1582 total mutants.