import zpools by guid on failover event

PR: https://github.com/truenas/middleware/pull/8506

This is already being done on SCALE HA so I'm doing the equivalent on 13. This prevents failover from not working when trying to import a zpool via name and 2 or more disks have the same zpool information stamped on the labels but they're actually different zpools.

This can be reproduced easily by creating zpool named "zz" and use 3 disks. Export the zpool and recreate the zpool using 2 of those 3 disks naming it the same. Try to failover and you'll see zpool give an error to import by numeric ID instead because of duplicate zpoolnames.