There's a TODO in persist.py:
# TODO scrape all event types in a similar way (writs should be consolidated in)
# Types should mirror the values from the HTML table headers, HR/ER/SE/etc.
I think this is solid, it might require bundling the hearings array up in to an "events" array? And populating hearing/event type dicts with a "type" field like HR/SE/ER.