block: switch to atomic_t for request references

refcount_t is not as expensive as it used to be, but it's still more
expensive than the io_uring method of using atomic_t and just checking
for potential over/underflow.

This borrows that same implementation, which in turn is based on the
mm implementation from Linus.

Reviewed-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

include/linux/blk-mq.h

@@ -139,7 +139,7 @@ struct request {
 	unsigned short ioprio;
 
 	enum mq_rq_state state;
-	refcount_t ref;
+	atomic_t ref;
 
 	unsigned long deadline;