Refs #34394 - trigger dynflow restart when DB restarts

When PostgreSQL is restarted (or worse, unavailable for a while),
Dynflow doesn't always cope well with this fact and sometimes ends in a
bad state, not able to recover itself.

This is clearly not fixing the underlying Dynflow issue, but at least
lets restart Dynflow in the cases we *know* the DB dropped connections.

manifests/dynflow/worker.pp

@@ -51,6 +51,8 @@
       enable    => true,
       subscribe => Class['foreman::database'],
     }
+
+    Anchor <| title == 'postgresql::server::service::end' |> ~> Service[$service]
   } else {
     service { $service:
       ensure => stopped,