This is mostly a tracker for me to review the details of the implementation of 791777d. I will update this issue with what I find.

Generally the commit was a good choice because previously there were issues with the test data sources and these needed to be recreated. It's just that there are some pitfalls in academic coding that I would like to avoid.

TODO:

• Review commit
• Change .xz archives to plain text
• My understanding is that the data is randomly generated, but we should include a script to generate it to avoid the same providence issue as before