The current behavior of run_tests --module is to playback all tests that have recordings and record all tests that do not (and are not live only). When I have to re-record networking tests, often there are a few tests that are very slow and block. During that downtime, I would love to replay the tests that have already been re-recorded and identify which, if any, fail to playback. If I do this by running the module, I'll inevitably have two processes trying to record the same test, and nothing good comes of that.
Consider adding a --playback-only flag to run_tests that will only playback tests, and skips any tests that are live_only (obviously) or aren't but don't have a recording file.
...and the summary of which tests fail at the end of the run would be super super super helpful.