[cassandra] Alert when a server is down

Today, the failure of a cassandra node is not detected by any of our monitoring systems.

It can be done in several ways, both have the pro and cons so they can be deployed both, we'll state later if they need to be kept both:

• Update the jmx-exporter configuration to export the org.apache.cassandra.net:type=FailureDetector DownEndPointCount and UpEndPointCount mbeans and declare an alert when there is some movements
• Monitor the cassandra ports via icinga as usual:
  • 7000/internode communication
  • 7070/jmx exporter
  • 7199/jmx
  • 9042/native transport