Make new endpoints for parsing content from PACER

[mlissner]

The RECAP server currently does the following:

New content:

• Receives uploads from the extensions.
• Parses the uploads and merges them into the database.
• Uploads the content to the Internet Archive as a primary data store.

Querying:

• Reports to the extensions about which items from a docket sheet are available on RECAP.
• Reports to the extensions about which dockets are available on RECAP, including their last upload date.

All of this functionality needs to get migrated to the CL server.

[mlissner]

• [x] Add authentication to the endpoint using a global RECAP user (use a migration to get this user in place).
• [x] Add stats to the endpoints so we can count per-user/date usage like elsewhere.
• [x] Normalize court id fields so that the API ingests and returns PACER IDs, not CL court IDs.
• [x] Add a parameter that links the processing queue items back to the dockets or documents they produce.
• [x] Update documentation (remembering that we have old docs for the original recap server)
• [x] Add validation rule that if upload_type is PDF, then we have to have a pacer_doc_id value.
• [x] Sort out a plan for Internet Archive updating.
  • [x] What if the SHA1 of an uploaded item differs from existing item? Replace on IA? How?

[mlissner]

This is a very big one to be marking complete. Hot damn, we have a functional and documented RECAP API on courtlistener!