scsi: core: Make sure that targets outlive devices

This commit prevents that the following sequence triggers a kernel crash:

 - Deletion of a SCSI device is requested via sysfs. Device removal takes
   some time because blk_cleanup_queue() is waiting for the SCSI error
   handler.

 - The SCSI target associated with that SCSI device is removed.

 - scsi_remove_target() returns and its caller frees the resources
   associated with the SCSI target.

 - The error handler makes progress and invokes an LLD callback that
   dereferences the SCSI target pointer.

Link: https://lore.kernel.org/r/20220728221851.1822295-2-bvanassche@acm.org
Cc: Christoph Hellwig <hch@lst.de>
Cc: Mike Christie <michael.christie@oracle.com>
Cc: Hannes Reinecke <hare@suse.de>
Cc: John Garry <john.garry@huawei.com>
Cc: Li Zhijian <lizhijian@fujitsu.com>
Reported-by: Mike Christie <michael.christie@oracle.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>