That sounds like it works, but shouldn't flaky tests be detected by frequent automatic test suite runs?
I've heard google does something fancier where they take a test, run it a bunch after it fails to check if it's failed
I think the system at work only runs each test a couple times before giving up and marking it failed