insights: spike prototype compute insights over time

Created by: coury-clark

This issue is to experiment with compute (mapping) insights over time.

Findings:

1. The query parser does not reconstruct compute queries properly (the content field is lost and therefore breaks) after adding fork or archive
2. The ComputeText response doesn't include repository metadata (or file, commit, etc). We would need to capture this metadata (even though the user could specify a per repo mapping in the query itself) so that insights can store the data with repository level cardinality.