Cinder wait for DB writes on reads

Because we deploy the database in multi-master mode we can have cases
where a service writes something in the database and when another one
reads from the DB that data is not yet there.

This is a problem for Cinder, so we need to ensure that reads happen
after writes from other nodes are present.

The cinder operator already has this [1], but until jobs are using that
code we will be adding the configuration option manually.

[1]: openstack-k8s-operators/cinder-operator@b074794

lib/control-plane/openstackcontrolplane.yaml

@@ -30,6 +30,10 @@ spec:
         # Debug logs by default, jobs can override as needed.
         [DEFAULT]
         debug = true
+        # Necessary to prevent DB race conditions.
+        # Remove once jobs run b0747944394807862e4cdcfa7052f1f8d1febf94.
+        [database]
+        mysql_wsrep_sync_wait = 1
       cinderBackup:
         networkAttachments:
           - storage