Recycle old download files when available

The download log stores the parameters and filename of each request, so it can be used as a caching system.

Roughly speaking, if there is a download in the database with the same corpus + parameters as a new request, you can just log the same filename and skip generating a new file.

It would be good to also build in some expiration date, e.g. only do this if the original download was started within the last n days. While the old file can remain available to the user that requested it originally, new users should not receive possible outdated data. (Due to updates or the like.)