Periodic dumps of column datasets

[davenquinn]

We should add a process to create dumps of entire project-based column datasets, with all of their attendant metadata (e.g., strat names and other dictionaries) to Geopackage databases.

This will allow us to have a process for people to acquire and use analytical sets of Macrostrat information that are at a specific revision. This will improve the transparency and consistency of analytical processes, as people can use a "frozen" set of Macrostrat information, even as the core database is improved. This can also empower people to make their own Macrostrat compilations by following our data model.

If we add content-based hashing of rows to key tables (e.g., units, columns, etc.) we can even detect changes made outside of Macrostrat, for integration of changes into our core datasets if desired.