Part of #689.

For the record, the idea is that in future:

The tags are stored on gcs instead of within a resources folder of malariagen_data, this would enable us to have them more readily available to users/partners, to version and update them (as we do with site_filters, for instance) or just update them (as we do with the metadata) as they get refined with more data (and the notebook already shows that they can be improved), and to have the structure in place if and when a similar list of tags is generated for funestus (which definitely has inversions), stephensi and darlingi (which probably have some inversions as well).