When a host is down, playbook execution will fail

Need to document how to work around this. We don't want the user to just remove the host from the inventory (unless it's a permanent change), since it will be missing from `prometheus.yml`.

Edit: an example of what I just did to deploy fresh dashboards while an OSD host was down:
```
ansible-playbook -v -i rhs_demo playbook.yml -l ceph-grafana -t dashboards
```

Start by reviewing the playbook.yml command, the rhs_demo inventory, and the prometheus.yml reference in the issue. Document how to target ceph-grafana and the dashboards tag while a host is down, and clarify why removing the host is only appropriate for a permanent inventory change.

Written by the indexing model from the issue text.