The crosswalk table file is changed every time a contribution is made without guarantee that the data, already in it, didn't change.
The proposition is to have a separate file for each mapping that will be aggregated into the crosswalk.csv file with a simple script.

This suggestion will result with some data duplication but could be much easier to use and will be more adequate for machine readability.

Before opening a PR for it, I would like to hear your thoughts, comments or additions to this suggestion, the goal is to make the crosswalk table more stable.