Complete rewrite of database layer

[TheTechRobo]

First commit was from February, geez...

things still to do (off the top of my head):

• Actually implement recursion. This should be fairly easy, add an API endpoint to queue pages and do that from the client. Depth should already be handled.
• Polish dashboard
  • Reimplement a few helpers
• Add !ig/!unig aliases for reject + skip rules and their corresponding removal
• Workers should report disk space to the server
• Display a relation graph on the dashboard (this one can maybe wait)
• Find all the overengineered stuff :-)
  • Decide whether the job rules system is necessary, or if it should just be a list of accept/reject regexes where reject takes precedence. (Perhaps skip rules should continue to be separated from reject rules in that case, though.) I don't know how useful varying user agent is between different parts of a job (especially considering those different parts could simply be split into multiple jobs - queuing to other jobs will be considered supported behaviour).
    • Actually, I kind of like the idea of having a separate array for each type of rule. That would also make it much easier for users to reason about.
  • Probably remove claim locks. UNTIL_FINISHED is covered by job tags (just make a new one that only has the pipeline(s) you want) and I can't really see a use case for INDEFINITELY. The only thing that tags don't cover for UNTIL_FINISHED is being able to fine-tune exactly how many of each pipeline there will be for a job, if concurrency > 2. Given that it's not much code, I'm open to keeping it in there for awhile in case it is considered useful.