Improve backup_spec reliability by waiting for the runner pod to restart
Summary
Follow-up from #5306 (comment 1763085628)
So my assumption now is that the runners were failing to register, or too late.
We call
restart_gitlab_runner
before running the tests which tells the runner to re-register after the restore. Although, this is run by simply deleting the old runner pod. But we only wait for its deletion, not for the new runner pod to start and register.My assumption now is that we didn't have this failure in the past because the UI was much slower and would give time for the runner to register. Now that we do API calls, we get to this test faster. So it might still be worth adding a tolerance wait script there. WDYT?