Store scap sync-world history in a SQLite database

This gives us:

• Safe updates (the old code did in-place overwrites with no way to
  recover if, for example, the program is interrupted or ENOSPC
  occurs at just the wrong time).

• ID numbers for deployments, which will be needed with a future
  spiderpig interface to rollbacks.

• Easier to understand GC mechanism.

• A baseline for how other future databases might be handled, such as
  the database that spiderpig will need (to hold job history, active
  interactions, etc).

In the process, drop the history_log scap configuration option.
Flexibility can be re-added when actually needed. I will manually
clean up the old history.log file on deploy servers.