We noticed a withdrawal was skipped.
There were some io issues towards the bsc node around that time ( which might have caused the connection to be re-established) and due to the way the leader scans the eth chains and then subscribes to events, there is a window there too that can cause misses.
Another issue: #110
The only way now to fix this is by restarting the bridge leader with a rescanheight lower than the missed withdrawal ( which is only done after a user notifies there is a problem).
A possible solution is for the cosigners to notify the leader about it.