If first unmount fails, e.g. due to "device or resource is busy", we are
trying to unmount the filesystem the second time after force
kill/FUSE-abort (see 5f684a49 "wcfs: Server.stop: Make sure to remove
mount entry even if we had to use FUSE abort").

This way the caller of Server.stop should get an error only if that
second unmount fails, not on unmount-1 error, which should be considered
as internal to Server.stop implementation.

If we don't hide that unmount-1 error and raise it to the caller, from
outside it can confusingly look like "the server is successfully
stopped, but nevertheless we are raised with an error".