Systemd units for stopped pods get stuck as failed

[nightkr]

This causes systemd to (nearly) always report the system status as degraded. Can be cleared manually by running systemctl reset-failed $namespace-$pod-$container.service.