re-derive missing object histories from database snapshots

After one of the recent hard drive crashes, we lost a large portion of encyclopedia object histories (this was before we were doing offsite backups of this data). The information is not gone for good, however – it is in principle (largely) derivable from database snapshots.

A rough plan for how to re-generate the snapshots is:

1. create a dummy Noosphere instance
2. iterate over the database snapshots
   1. uncompress each snapshot
   2. load snapshot into database
   3. have Noosphere dump an XML version snapshot to disk for every object (this is an existing subroutine).
3. Replace missing and broken snapshots in live versions/ directory with "synthetic" snapshots from previous step.

There are two caveats to this plan:

• revision comments are lost forever; restored snapshots do not have them, as they are only stored in the original XML.
• multiple changes in the same day will be lost.

However, I think most of the desired effect will be attained with this method.

As a related project, the version history page should be improved so that missing intermediate versions are handled more elegantly (one should be able to diff between on-adjacent versions, or even arbitrary versions).