List privacy risks

[dsmits]

We need to be transparent about the potential privacy risks of verticox+ Let's use this issue to keep track.

• For cross validation we need to specify which records are used during training and testing. This can potentially be manipulated to retrieve person-level data
• Records need to be ordered by someone
• Aggregator algorithm currently needs to be run on outcome node

[dsmits]

Should take this from the paper once it's published