v0.183.56: add lock even when volume is active and detached

we should also add lock in scheduling instances when volume is active
but detached. This will prevent a race condition when we try to
schedule instances using an existing volume or when it failed to pass
healthcheck and reschedule containers using existing volume.