Add Telemetry metrics for task drift

[shanisagiv1]

We do not have visibility into how many customers are under provisioned. To get a better understanding of how many clusters are under provisioned, let's utilize event based telemetry to make all the Kibana instances report certain metrics on a 10 minute interval:

• The number of tasks that ran <= 30 seconds of their original runtime within the interval
• The number of tasks that ran > 30 seconds of their original runtime within the interval
• The number of polling cycles within the interval
• The number of polling cycles where all workers were filled within the interval

Definition of Done

• [ ] All metrics are reported
• [ ] All metrics reset after reporting them
• [ ] Unit and functional tests added
• [ ] Telemetry cluster receiving data

[elasticmachine]

Pinging @elastic/response-ops (Team:ResponseOps)