Stop using CouchDB for ignore patterns

[hannahwhy]

This is a more focused portion of #132.

Ignore patterns are just JSON files and can be read from disk every time an ignore set is requested. This ensures that the data in the local copy of the ArchiveBot repo will be used, which means ignore set deployment becomes just git pull.

However, for Historical Reasons, ignore sets are read from a CouchDB database and therefore need to be updated. There was an agent that watched ignore sets and did this updating automatically, but it or some of its dependencies no longer function.

This could be fixed, but I think the Better Fix is to just cut out the middleman.

[JustAnotherArchivist]

May I suggest renaming this issue to "Remove ignore patterns from CouchDB" (or maybe "ignore sets", to be really precise)? The current title can be misinterpreted as "retrieve ignore patterns from CouchDB"...