Handle IEA extended world energy balances

[khaeru]

Closes #25.

The code supports the same uses as the more recent https://github.com/SiddharthJoshi-Git/IEA_activity_calib_SSP.

How to review

• Check out the branch.
• Attempt to follow the docs instructions, using the data with a full-scale database (not included).

PR checklist

• Complete the code
  • Add fuzzed data for testing.
• Continuous integration checks all ✅
• Add or expand tests; coverage checks both ✅
• Add, expand, or update documentation.
• Update doc/whatsnew.

[khaeru]

Just a note that this should probably be using PyArrow, e.g. https://francoismichonneau.net/2022/10/import-big-csv/

[codecov]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (fdfb4c9) 74.9% compared to head (1988bc1) 74.5%.

Additional details and impacted files

@@           Coverage Diff           @@
##            main     #75     +/-   ##
=======================================
- Coverage   74.9%   74.5%   -0.4%     
=======================================
  Files         91      93      +2     
  Lines       6084    6204    +120     
=======================================
+ Hits        4558    4625     +67     
- Misses      1526    1579     +53     

Files	Coverage Δ
message_ix_models/project/ssp/structure.py	100.0% <100.0%> (ø)
message_ix_models/testing/cli.py	88.8% <100.0%> (+3.1%)	⬆️
message_ix_models/tests/project/test_ssp.py	100.0% <100.0%> (ø)
message_ix_models/tests/tools/iea/test_web.py	100.0% <100.0%> (ø)
message_ix_models/tools/iea/web.py	100.0% <100.0%> (ø)
message_ix_models/util/sdmx.py	97.0% <90.0%> (-0.9%)	⬇️

... and 2 files with indirect coverage changes

[khaeru]

The codecov/project check failure here appears to be erroneous. Codecov shows missing lines in .model.snapshot, but the job itself shows 100% coverage in the same file.