fix(cron): close lock_fd on failed flock to prevent fd leak

[alireza78a]

tick() opens lock_fd before the try block, so if fcntl.flock() raises BlockingIOError, the except clause returns without closing it. In a long-running gateway process where concurrent tick attempts are normal, this leaks a file descriptor on every skipped tick.

Fix: initialize lock_fd = None before the try block and close it in the except clause if it was opened.

[teknium1]

Merged in fec8a0d 🎉

Clean catch — in a long-running gateway ticking every minute, this would leak one fd per concurrent tick attempt. Thanks @alireza78a!