Tweaks to reduce billing costs

[ItsMeBrianD]

The majority of our (admittedly small) backblaze bill is coming from S3 Scan operations, which is caused by building pages

We can probably be smarter about how we are building the pages based on the queries that were actually run. This also makes it easier to remove datasets that are no longer used (e.g. Player Stats is no longer updating), because they will no longer update.

We also want to make sure that we aren't storing versions of the data files in backblaze because we have tar archives for previous versions of the datasets available on the archives page. This will help keep our storage level much flatter than it would be otherwise