Put data-raw/csv/ under version control and include traitsview as the primary table

[dlebauer]

Following this discussion, I initially thought it would make sense to add traits and yields table CSVs and use these to construct traitsview in data-raw/make-data.R.

On further consideration, a better solution is to maintain a single version of traitsview that includes:

• all records in traits
• all records in yields
• additional fields (if any) from traits and yields that are not in the traitsview table

The goal is to have a single primary file to edit, with supporting metadata and keys (as is currently the case) that link to metadata tables that can also be edited. Using the unified view rather than the original traits and yields tables will make editing easier and less error prone (because humans will be able to read the natural foreign keys as they are editing.