Task tracking changes flaky summary counting algorithm

Update top flaky tests measurement with additional field allowing seeing if the failures are Health check only. This would allow for distinguishing if potential  flakiness of  test due to CI. So far we do not have algorithm allowing for automatic classification if the root of the flakiness is in the test itself or the running environment (CI). The triage (classification) is done manually by checking logs. Calculated, refined numbers (splitting the fail counter to fails in integration vs fail in health check runs) would provide more insight obtained at one click.