Doc

Monitor test durations: Flex test timeouts to allow long running tests. We should have a long timeout which will kill stuck tests eventually. Tests should include durations for pod becoming ready, cluster becoming ready and we need to check these durations. Between two releases time to cluster readiness may change significantly and making this time longer should be considered as a regression.

• Single test duration should be limited to 1h max (or maybe even 45min), if it’s longer split it into two or more tests. Reason: if the test fails somewhere at the end and you make some changes to fix it you need to wait a long time to check if it will pass or not.