Add page count to crawl model

[tw4l]

One of the consequences of adding pages from uploads to the database in the backend work for public collections (added in #2198) is that we now have pages read into the database for both crawls and uploads, but in various places throughout the application (e.g. archived items list, list of archived items in a collection), we report the number of pages for crawls but not uploads. This is because what we currently use for a crawl's page count (crawl.stats.done) only exists for crawls.

Now that we have pages in the database for crawls and uploads, we should count these documents per archived item and store the count in the crawls model in a way that is consistent for all archived items, and then use that new count across the frontend where we list page count for archived items.

This will likely require a migration to backfill data as well.