[Bug Report + Patch] File descriptor leak when io_method=io_uring

Hello Hackers,

I noticed a file descriptor leak when running PostgreSQL 18 with
io_method=io_uring. As far as I could tell, it only triggers when
the server restarts due to one of the backends being killed.

How to reproduce the issue:
- Setup a server configured with io_uring
- Check the number of open file descriptors:
ls -la "/proc/$(head -1 $PGDATA/postmaster.pid)/fd/" | grep "io_uring" | wc -l
- SIGKILL a backend to trigger a server restart:
kill -9 $(psql -XtA -U postgres -c "SELECT pid FROM pg_stat_activity WHERE backend_type = 'client backend' LIMIT 1")
- Check the number of open files descriptors again:
Expected: FD count remains the same.
Actual: FD count has doubled.

Tested on PostgreSQL 18.0, 18.1, 18.2, 18.3 and git master on the
following platforms:
- Ubuntu Server 24.04: Linux 6.8.0, liburing 2.5
- Exherbo Linux: Linux 6.16.12, liburing 2.12
- Fedora Linux: Linux 6.19.7, liburing 2.13-dev

From what I could gather, this happens because
pgaio_uring_shmem_init() in
src/backend/storage/aio/method_io_uring.c doesn't cleanup
the allocated resources on exit.

I've attached a patch which registers an on_shmem_exit() callback
to close the file descriptors on server exit. I took inspiration
from how src/backend/storage/aio/method_worker.c handles cleanup.

Regards,

Lucas.