Set up system for automatic asset re-extraction

[jjnesbitt]

Following on from #1545

That PR is a manual solution to the problem of assets metadata becoming out of date, as the schema changes. It may be beneficial to automate this process, with a job that checks for assets that have an old schema version, and kicks off jobs to re-extract those assets.

However, if this re-extraction is mostly affecting old assets (i.e. if the schema will rarely change going forward), then I don't think it justifies the engineering cost, and will just get in the way of other efforts. The schema was changing at a much higher rate in the past, and so running re-extraction on those old assets may take care of >90% of the issue for the foreseeable future.