Add a general-config feature to add all countries to RegionCodeList

[danielhuppmann]

This PR adds a simple way to add all countries to a RegionCodeList directly from the pycountry package, using a "general-config" yaml file.

This implements the first idea of #261.

[danielhuppmann]

When looking at the pycountry package in more detail for the SSP review process, I realized that it uses some country-names that are not consistent with the use in our community. So I guess that we will have to hard-code the country-names and ISO3-codes after all...

[danielhuppmann]

After looking at the pycountry package some more, I suggest (and implemented) a mapping of "country-name-overrides" where we explicitly depart from pycountry standards. Seems like a preferable option than to duplicate an entire list of country names and ISO2/ISO3 codes...

The model-registration xlsx template has been updated accordingly.

[phackstock]

Looks good to me.
The only comment I'd have is on the use of the DataStructureConfig.
Right now its just a wrapper around a dictionary. I think you could save a lot of manual checking, for now and for the future, if you completely defined the structure of the DataStructureConfig and its dimension specific settings.
My inline comments are regarding that.

[danielhuppmann]

Implemented the suggested changes, @phackstock, towards nested pydantic classes. And I also added some basic documentation - I'm planning to add more documentation in a follow-up PR, then we can do a release 0.11

[phackstock]

Looks good to me.
The pydantic config classes make me very happy, thank you :)
One question that came to my mind was if it would be easier to just work with ISO codes directly as the region names. Might be easier for modeling teams to align their outputs.

[danielhuppmann]

One question that came to my mind was if it would be easier to just work with ISO codes directly as the region names. Might be easier for modeling teams to align their outputs.

Right, @willu47 made a similar point here: OSeMOSYS/osemosys2iamc#33 (comment)

But I don't really like the idea for several reasons:

• ISO3 codes are not really user-friendly - it's difficult to get a quick intuition in a tool like Artelys' FutureSight or the Scenario Explorer, so we would probably invent some "conversion tool" again
• ISO3 codes are not fully universal either, e.g., Kosovo doesn't have one
• Using ISO3 codes in native-country naming also has issues, for example REMIND 3.2 uses "UKI" internally for "United Kingom and Ireland" - this can be very confusing when users try an ISO3-mapping on that... (see here).

So I think it's better to nudge modelers to use intelligible names instead of abbreviations.

Nonetheless, we should add a simple ISO2/ISO3 mapping tool to the shortened country names added in this PR (and write documentation for it).

[phackstock]

But I don't really like the idea for several reasons:

Ah right, good points, it's never as simple as you think ...

Nonetheless, we should add a simple ISO2/ISO3 mapping tool to the shortened country names added in this PR (and write documentation for it).

Good point, that would be surely appreciated. In the context of NGFS, the users were specifically asking for the country level data to be reported in ISO3 so that would come in handy there.