When pausing hacluster and principal unit, haproxy still references stopped unit in backends

I believe the openstack api layer is the correct place for this bug.

We are seeing issues with paused units still being referenced by haproxy from other HA peer units.

As an example, deploy 3 units of designate on focal-ussuri with https endpoints (via vault:certificates relation)
Then 'run-action pause' against one of the designate units and it's hacluster-designate unit.

Then every 3rd or so hit to 'openstack zone list --all' returns:
Failed to contact the endpoint at https://designate.mysite.com:9001 for discovery. Fallback to using that endpoint as the base url.
NotFound

This seems to be related to there being no healthcheck routine within the haproxy backend configuration templates.