Deduplicate generated alerts in HA setup

Removes the instance-specific 'monitor' label set on alerts, so that
multiple Prometheus instances in a HA setup all create the same
alerts.
1 job for master in 4 minutes and 56 seconds (queued for 2 seconds)
Status Job ID Name Coverage
  Test
passed #8901
ai3
run_base_test

00:04:56