Is your improvement request related to a feature? Please describe (? if you like this request)

When the user creates a volume and makes a backup, there is a mount point in IM-R for the replica to do the backup.
If the user unsets the backup target, longhorn won't unmount the mount point in IM-R and might introduce system instability such as high CPU loading when the backup server is gone.

We have tickets related
#5391: every time when LH tries to mount the mount point, it will unmount other mount points first and then mount the current setting Target.
#5655: backup-target-controller will unmount mount points in longhorn-manager pods when the URL is unset.

We would like to handle IM-R pod mount points left issue later when we implement multiple backup target support feature in #5411

Describe the solution you'd like

When the user unset the backup target, the mount point in IM-R pod should also be unmounted.

Describe alternatives you've considered

A clear and concise description of any alternative solutions or features you've considered.

Additional context

We might need to consider adding a new grpc service in IM-R for IM-R to do the cleanup in the pod.
Currently, there are only ProxyEngineService and ProcessManagerService these two services which are not appropriate for this general task. (see #5655)