[discussion] fail/flake detection threshold
Currently when we encounter fail, we run the tests again (1x) and decide if it's fail or flake.
I believe it's pretty risky, because flake, which accidentally flakes two times (2x) in the row will lead to marking thest as a fail.
I would like to suggest to move default retry value in case of fail to five times (5x), so we can be sure, that it's really failure.
Q1: For your use case, would it make sense to have a 5x as a default? (accounting for also having !66 (merged) in)
Q2: Do you find useful to be able adjust the number of fail/flake testcase repetitions via command line?