Ensure that the cumulative chaos metrics are not dependent on the presence of chaosengine/chaoseresult resources on the cluster

[ksatchit]

The total run/pass/fail metrics coming from the exporter today are dependent on the presence of the chaosengine and chaosresult resources in the cluster. In other words, they are providing state for the exporter. Upon periodic polling intervals OR upon startup - the exporter scans for these CRs and calculates the metric values exposed over :8080/metrics.

However, with logging integration having been achieved, we no longer plan to keep these CRs on the cluster. They will be auto-cleaned up. This will cause the exporter to throw up incorrect values for these cumulative metrics.

We need alternative state-management. Potential approaches include usage of textfile collectors, querying from control plane (db), or moving to the prometheus push-model.