PostgreSQL_ReplicaStaleXmin alerts

Since May 24th we've had alerts firing every few hours about PostgreSQL replica being stale, see here. I also noticed that similar alerts were firing on April 16th and 17th

Should an incident be opened for this? if not, should we adjust the alert threshold? here's the alert definition: https://gitlab.com/gitlab-com/runbooks/blob/master/rules/postgresqls.yml#L162

@yguo @abrandl @cshobe @amanikavasagam @Finotto Could one of the DBREs investigate this please? This will trigger an alert (wake up people on call)