Drop redundant pending increment from wakeup-read submission.

After "Handle short io_uring submissions" (b57476a) moved the `selector->pending += 1` into `io_get_sqe()`, the manual increment in `select_internal_without_gvl` introduced by #166 became a double-count.  The duplicated count left `pending > 0` after `io_uring_wait_cqe_timeout` drained the SQ internally, sending the next `io_uring_submit_flush` into a busy loop (submit returns 0 because the ring is empty, but `pending > 0` keeps the new while-loop spinning).

Co-authored-by: Cursor <cursoragent@cursor.com>

Author: samuel-williams-shopify

ext/io/event/selector/uring.c

@@ -1139,7 +1139,6 @@ int select_internal_without_gvl(struct select_arguments *arguments) {
 		io_uring_prep_read(sqe, IO_Event_Interrupt_descriptor(&selector->interrupt), &selector->wakeup_value, sizeof(selector->wakeup_value), 0);
 		io_uring_sqe_set_data(sqe, &selector->interrupt);
 		selector->wakeup_registered = 1;
-		selector->pending += 1;
 	}
 
 	io_uring_submit_flush(selector);