[scorecards] have more robust data pipeline for scorecards data

[struan]

At the moment this is a very manual process which involves running a script on the marking site and then moving the right files into the data directory of the scorecards site. It would be less error prone if much of this was automated and possibly we stored the data somewhere that it could be versioned/tagged so we could keep track of changes/history of updates.

The easiest might be a private git repo with the files that scorecards fetches from and marking uploads to.