Ensure /var/log/journal exists as soon as possible

It might happen the /var/log/journal directory doesn't exist on a host,
especially if this host is a VM (the image creation usually involves
deep cleaning of the logs).

This absence might lead to log loss with the default journald
configuration, which uses the "auto" Storage.
This "auto" means that:
- if /var/log/journal exists, journald will use it, and it will be
persistent
- if /var/log/journal doesn't exist, journald will use a volatile
location, /run/log/journal, which is dropped upon system reboot

Since logs are important, and "old" logs might be useful after a reboot,
it's better to ensure we have persistent storage for journald.

Related-Bug: #1856278
Change-Id: I93dcc57aff63b91dab475b0c114b278324434e41

common/services/role.role.j2.yaml

@@ -393,6 +393,14 @@ outputs:
           - {get_attr: [DockerPuppetTasks, value]}
       host_prep_tasks:
         list_concat:
+          - - name: Ensure /var/log/journal exists
+              file:
+                path: /var/log/journal
+                state: directory
+                mode: 0750
+                owner: root
+                group: root
+                setype: var_log_t
           - - name: Run firewall role
               include_role:
                 name: tripleo-firewall