Timing-related tests such as the spec and prose tests for CSOT tend to be far more flaky on Mac and Windows OSes. Per this PR discussion, to reduce noise from failing tests, it may be prudent to add a new flag for timing-related tests (maybe in the form of an environment variable + mtest configuration) and just skip those tests on those OSes. While this may reduce test coverage, timing failures rarely represent actual bugs in driver code, so these recurrent failures are not particularly helpful and can be difficult/time-consuming to fix.