Add high availability support to the deployment architecture

At the moment the configuration files are read via file://. If the filesystem dies, the entire deployment configuration is lost (unless backed up). There is still an unavailability window where deployment updates couldn't be done in this condition.

What would make more sense is 2 or 3 nodes which replicate the /etc/kolla directory contents to a backend that was a multi-node data store. Since Ansible gets all state of the current system during the runtime, it doesn't need any more storage then a) whether a deployment operation is in progress and b) the configuration of the deployment.

I propose modifying the merge file script to read from configurable data stores. We would also add an Ansible script which read the contents of /etc/kolla and loaded it into a configurable data store. Finally there would be a "deployment_active" flag which would prevent new deployments unless the operator manually killed the flag (because they were aware of a lights out situation during a deploy which would result in the deployment_active flag not being removed from the data store.

I'd like the loading versioned, so for example there is a:
/0/etc/kolla representing the first load of the configuration
/1/etc/kolla representing the second load of the configuration
etc...