VirtualDomain VMs aren't migrated during node shutdown.

### Steps to reproduce

* configure a two node cluster
* configure a VirtualDomain resource active on one node
* reboot the node hosting VM

### Expected behavior

The resource will be migrated to the other node using `migrate_from` and `migrate_to` RA actions.

### Actual behavior

The RA cannot query VM state and the `migrate_from` operation times out. (Alternatively, VM is shut off by systemd shutdown transaction before VirtualDomain RA can act on it.)

Technical details and proposal is in the referenced patch.

Start with the VirtualDomain resource and its resource-agent handling of the migrate_from and migrate_to actions. Reproduce the two-node shutdown scenario, then investigate why VM state cannot be queried or why systemd shuts the VM off before the agent acts. Done means migration completes during node shutdown without the migrate_from operation timing out.

Written by the indexing model from the issue text.