Repair failed mount of /var

We have some machines which are experiencing a boot problem with the following error message quite regularly.

mount: /var: cannot mount none read-only.

We have no clue why aufs is not mounting correctly, but a manual systemctl start var.mount works every time.

This installs a repair script which just starts the mount again on a failure. This can lead to an infinite loop, but my tests have shown two calls to repair-var.service until /var is mounted correctly.

I added systemd-resolved because it fails, when var.mount fails the first time and without it chef-client does not find its server.