insights: retry policy for errored queries is too narrow

Created by: coury-clark

Currently, the query queue handler provides retries for failed Sourcegraph queries but is configured to retry too narrow, causing queries to erroneously fail during routine network outages in the Sourcegraph cluster.

These queries are meant to be long-running and should be resilient to even severe network outages. Probably these values should be something like 30-minute retry interval and 100 max retries.