The connector does not presently produce any heartbeat events except when we emit a change event which doesn't help the overall problem with flushing offset state when there is low change event activity or none at all.

We should force heartbeat events in the main event processing loop when we update the offset's SCN as well as when we perform a commit to guarantee that during low or no change event activity, actions seen by the connector can see make sure offset state gets replicated to offset storage appropriately to avoid any stale state creating issues upon restart.