I reviewed the python and gave Joel a bunch of notes. tl;dr was that everything basically looked good, with one potential big speedup (or not), and some ideas for making the code easier to understand.

Any objection to my moving the feedback from a google doc to a new Phab task (so I don't lose it/can treat it as backlog?)

@JAufrecht: Bringing the notes into phab makes sense to me. Go for it.

My SQL skills are "intermediate", so I don't feel capable of doing a thorough review. I will say that the code is well-commented, and all the parts that I did understand (more than half) made perfect sense.

Personally, I would like to see most if not all of the code moved into python. The calls to export tables as CSV would (mostly?) be one-liners, and the munging could be there as well. In some cases, the python would be clearer, and in other cases it would be less clear, but overall I would call it a net win to no longer have raw SQL scripts as part of the process.

Done as soon as Joel stores Kevin's email feedback in a permanent, public location. Actual work to address anything is new tasks.