Stackdriver can overlap different logs-based metrics

[bnelson729]

We had two logs-based metrics named ingestion-run-time and ingestionRunTime with different types and descriptions. The stackdriver exporter metrics endpoint was returning a 500 error saying that the help text didn't get what is was expecting to find. I believe the stackdriver exporter was trying to map those two different log-based metrics onto the same prometheus metric name.

We have since renamed our metrics to not overlap, but it would help if the error handling was better. There was no mention of the issue in the container logs. The error was only displayed on the /metrics endpoint.