End to end integration tests for `sources` about li

[jzohrab]

Adding some notes in a google doc to think of some test scenarios.

[jzohrab]

Every test scenario in that google doc has been added except for this:

Every date in the live prod cache should be scrapable, and not return any errors. We will iterate through all dates in the cache (potentially even every date/time, which indicates a data set), and the corresponding scraper should return data. Some crawlers use an array of “subregion names” (e.g. county names) to create URLs, but then change that list over time. That would result in cache misses, which must never occur.

I still think this is necessary for successful regeneration and system/data stability.