Additional datasets for different `geo_type`'s (moved to epiprocess)

[kenmawer]

This issue was moved from epiprocess to epipredict due to the latter loading the datasets.

There are five different geo_type's recognized, but we only have "county" and "state". We don't have geo_type's "hrr", "hhs", "nation" and "custom".

• "county" (not country): 5-digit FIPS code in the US
• "hrr": USA hospital referral region, from 1 to 457
• "state": (USA) State, including Puerto Rico ("pr") and Washington D.C. ("dc"), abbreviated by a two-letter lowercase state code such as "wa" for Washington (state) and "tx" for Texas
• "hhs": U.S. HHS region, coded by number between 1 and 10
• "nation": country, coded with a lowercase ISO 31661 alpha-2 country code, such as "us" for the United States and "ca" for Canada (not California in this context)
• "custom": anything else, for example health regions in BC; one can use shorthand codes for these as well such as "vch" for Vancouver Coastal Health in this example. Also, states in Australia are here, even with a predefined "state" geo_type (should perhaps be made unambiguous).

Here are a few things to note:

• There are no name conflicts with two-letter codes with Canadian provinces with the two-letter codes of US states, meaning that "state" could also accommodate provinces
• If we were to include other countries (including Canada) and/or other type of geo_type's, one would need to consider the context. Here are two examples: For example, "wa" can refer to Washington state in the US or Western Australia state in Australia, both of which use "wa" as a code and are called "states."

Here are the ones that are not here yet:

• "hrr"
• "hhs"
• "nation"
• "custom"