Automatically run cbcollectinfo when node fails unexpectedly

We should automatically capture collectinfo from a node (Alk prefers 'all' nodes in a non-obtrusive manner) in all node down/auto-failover scenario, this would ensure we have diagnostic logs captured immediately after such events preventing the 'logs rolled over, we don't know why it failed' scenarios that support repeatedly encounters in the field.

This was also suggested by customers as well.