When installing a cluster, in case of issue, the user has the possibility to restart the cluster installation (without the need of re-inserting the installation configuration detail).

Implementationd details

To support that, the backend should restart the `agent.service` (and cleanup the assisted-installer container). The trigger condition may be activated by the UI by writing a sentil file in a shared volume, monitored by a .path systemd unit

Current workaround
Reboot the entire rendezvous node to restart a fresh installation from scratch