Use `targets` for data pipeline management

[aridyckovsky]

Problem Splitting our analysis by R notebooks makes sense only for exploration and storytelling. When we have an actual data pipeline figured out, we ought to have it in a real pipeline that we can reproduce without multiple notebooks of many chunks.

Solution Use drake (link: https://github.com/ropensci/drake) to handle our data pipeline.

[aridyckovsky]

After more review it appears drake was superseded by https://docs.ropensci.org/targets/