RFE: Ability to query cpu seconds metrics

[afcollins]

I would like to see the "cpu seconds" counter metrics (added in kube-burner#454) trending over time.

Having all queries from that PR would be good:

* nodeCPUSeconds-Workers
* nodeCPUSeconds-Masters
* nodeCPUSeconds-Infra
* cgroupCPUSeconds-Workers
* cgroupCPUSeconds-Masters
* cgroupCPUSeconds-Infra
* cgroupCPUSeconds-namespaces

It will be helpful having the ability to query a single metric (i.e. a specific namespace of the metric cgroupCPUSeconds-namespaces) AND being able to query a metric and "group by" a field in the metric (i.e. group by "namespace" of cgroupCPUSeconds-namespaces).
The former allows us to track specific components, and the latter allows us to find issues in components we aren't explicitly tracking.

I am able to achieve this grouping vs filtering at the elasticsearch query level through touchstone-comparison (OK, maybe that's not a pure elasticsearch query in that case), like the following:

      {
        "filter": {
          "metricName.keyword": "nodeCPU-AggregatedWorkers"
        },
        "buckets": [
          "labels.mode.keyword"
        ],
...
      },