Add support for Ceph monitoring

This change adds support for Ceph monitoring:
- service.monitoring.cluster_stats is applied to monitoring nodes for
  collecting cluster-wide metrics (through the Ceph CLI).
- service.monitoring.node_stats is applied to the Ceph nodes for
  collecting monitor and OSD metrics (through Unix sockets).

Because Telegraf runs as a container on the monitoring nodes and
requires a working Ceph client configuration, this  change also adds
support for deploying Ceph client in container mode.

Change-Id: If7359aca34a350f2c8ee2251bbe8a85314550a45
12 files changed