Fix openstack CI (#714)

* Wait for the router and registry pods to come up

The end to end openstack CI started failing intermittently, because
the openshift-ansible finished without waiting for the router to come
up. If we waited with the validation, it would succeed.

So this adds two checks that verify that the registry and router are
actually running or failed before considering the installation done.

* Print diagnostic info about the cluster

oc get nodes, oc status and oc get all can all be useful in
determining why the test failed.

* Add missing `fi`