Consider version controlled ETL scripts for data resources

[seandavi]

Scripts that can grab data, transform it (including pulling in necessary curation), do testing, and then load to final locations will increase reproducibility and decrease the burden of updating resources when the source changes.