Add retry logic to zlonk on database recovery time outs

Per https://gitlab.com/gitlab-data/analytics/-/issues/10408, we've had two instances where the database recovery failed and was timed out. SRE On-Call, following the runbook, resolved the issue by recreating the clone. We should do that automatically yo avoid disturbing SRE and avoid disrrupting the data flow.