Additionally to getting baselines for categories as part of T383138.

• Get data on % low-level categories (filter out root categories) that are added month over month
• Get data on newly created one page categories

Probable approach:

• compile a list of hidden, maintenance, and non-diffusing categories
• script to listen to the event stream, then use api to gather non-hidden/maintenance categories for each new UW upload (and record it)
• keep file uploads that have edit tag ID = 12, i.e., UW
• any non-hidden/maintenance category that has a wikidata item counts as a non-low-level category? or is there a way of telling if a category has no parent?
• instead of low-level categories, we might look at a measure like the average depth in the tree