I can’t find anything specific to what you’ve seen, but there have been a number of race-condition bugs in rootless podman where containers are running multiple processes that are similar to the problem you’ve seen.
The recommendation from podman is “don’t do that” or at least have a top-level process that manages its own children properly during shutdowns and crashes. There are several patches going back a while fixing ways this problem can occur. If you’re using a recent VyOS, it should be running a recent podman release (4.9.x) so those fixes will be included already.
There are other problems that can cause unmounting overlays to fail - usually these would be actual device problems, like unresponsive block devices or NFS.
What appeared to be in the overlay merged dir?
Was there anything else in the logs which might indicate an unexpected failure/restart of the container, like an OOM-kill or process crash?
Anything in the logs indicating problems with container stop and fs cleanup?
Any other problems with storage?