Add food affordability data

[pabloarosado]

Add steps for World Bank's Food prices for nutrition dataset.
Also, I added a small improvement of etl.harmonize: To include the region code as another alias.

[paarriagadap]

Looks good! I mostly added comments in the metadata to change the units of the monetary measures in constant international-$

[paarriagadap]

The assignment of (WB) seems inconsistent between the WB income groups: Low-income countries has it, but not High income or Upper-middle or Lower-middle income

[pabloarosado]

Good catch! By the way, I'm assuming that their income groups are consistent (and up-to-date) in all World Bank datasets (and hence consistent with OWID), let me know if that's not a good assumption.

[paarriagadap]

I assume that as well, but it will depend on when that dataset was released. If it's sufficiently old it can have a previous definition of income groups. I see they change every year.

[paarriagadap]

Now that you say it, it should be that groups are dynamically changing per year, so that would be consistent. I don't know that with 100% certainty.

[pabloarosado]

If they assume a definition of income groups that changes over the years, then those income groups are inconsistent with ours. Because we assume the current definition of income groups. But for this dataset (with just a few years of data) I suppose it should not matter much.

[paarriagadap]

Right, I wasn't sure about that. I would assume theirs is dynamic, but I don't know

[paarriagadap]

In general we are using the hyphen only for lower-middle or upper-middle. I don't think I have seen high"-"income or low"-"income or lower-middle"-"income much.

[pabloarosado]

This is currently how we name income groups:

    MAPPING_CLASSIFICATION = {
        "..": np.nan,  # no available classification for country-year (maybe country didn't exist yet/anymore)
        "L": "Low-income countries",
        "H": "High-income countries",
        "UM": "Upper-middle-income countries",
        "LM": "Lower-middle-income countries",
        "LM*": "Lower-middle-income countries",
    }

For consistency, those are the names we should use. We could consider changing them (but it would be a significant refactor, similar to the one of East Timor).

[paarriagadap]

It's weird that they use these units here. It won't be of much use to compare countries, unless only the year 2017 is shown.

[paarriagadap]

I see they use the ratio between this and the cost of an energy sufficient diet below (and many other cost afterward), which is not wrong just because the latter only has data for 2017

[pabloarosado]

This one is unclear to me. Currently in our chart we assume it's 2017$ (like all other variables), but that may not be accurate. I'll merge for now and consider restricting the chart to 2017.