Flaky Test Detection Heuristic Improvements

[rohan-at-sentry]

We will do the following improvements

Better flake expiry (1 SP)

• Calculate and use the “flake percentage” (how often does this test fail on main) to determine when to expire a flake. A 100% flake would be expired after 10 passes
  • a 1% flake would be expired after 1000 passes

Error Message Parsing (3 SP)

• Use the work done for framework detection in [TA] Error message parsing #1487 to improve our flake normalization system
• Alternatively find and implement other approaches