Continuous parser

[filipsch]

Currently, the parser is incremental: it parses new packages, and stores the doc files in the database. Old packages that were parsed before, are not being re-parsed.
If we update the parsing logic, we want all pages and docs to benefit from these changes.
Proposal: put a system in place that reparses everything e.g. every month.

[filipsch]

To make sure that it doesn't start processing huge batches as soon as we start it up, we can do two checks in the first months:

• was the package update more than 27 days ago?
• Was the last package release on the same day of the month as we are now?
  That way, the jobs are going to be spread out more evenly until the updates have happened for all packages.