Add subnational GDP

[kavigupta]

Both present and historic

This paper presents a dataset that goes up to 2015. https://www.nature.com/articles/sdata20184

This paper refines the technique from the previous one, but only provides 2005 data: https://www.nature.com/articles/s41597-022-01300-x#Sec17

I think it makes sense to use the first paper, since it has a more recent estimate, even if a bit out of date. It also has HDI estimates

OK the data seems very problematic. A bunch of subantional regions end up with 0 GDP, and there's also places that end up with GDP per capita PPP of 40 cents (Taoudénit, Mali) or unrealistically high values (2.4 million for Isla Congreso, Spain).