LSIF: Replace node-resque with bull

Created by: efritz

This PR replaces our dependency on node-resque with bull. I made a bad initial choice that caused a bit more friction than necessary when splitting the server and worker.

The latter library is 12x as popular on npm and seems more active in general. This new library has additional features we'll want if we end up sticking with a redis-based queue for work (job progress reporting, priority queues, cron scheduling). This simplifies our setup code a great deal and has working type declarations we don't need to declare ourselves.

I'm doing this now as a scheduled enqueue is the easiest way for us to do work on a schedule (required for RFC 43) while ensuring that (1) doesn't interfere with queries or ongoing conversion, and (2) isn't done by multiple processes.