Miscellaneous improvements for 2024-W14

[khaeru]

• Migrate handing code for 4 sources of exogenous data from message_data.
  • Global Energy Assessment project.
    • Add fuzzed data for testing.
  • Global Fuel Economy Initiative.
    • Add fuzzed data for testing.
  • IEA Energy Efficiency Indicators.
    • For this source, no fuzzed data are added for testing. This is because the source uses a complicated, idiosyncratic file format. Duplicating this format is out of scope for the current PR. The lines are thus excluded from test coverage.
    • For the 2023-11 edition of these data, the IEA website at https://www.iea.org/data-and-statistics/data-product/energy-efficiency-indicators provides a "Demo and availability file", which we could use for testing purposes.
    • → Handle IEA EEI 2023-11 edition #164
  • SHAPE project.
    • Add fuzzed data for testing.
• Improve docs of all of the above.
• Add path_fallback(), generalized from a utility function used in message_data/MESSAGEix-Transport.

Housekeeping:

• Adjust MANIFEST.in after Migrate legacy reporting #159.
• Remove snapshot files also available from Zenodo; these were committed in Migrate legacy reporting #159.
• Use nschloe/action-cached-lfs-checkout to reduce Git LFS bandwidth usage.
  • As we accumulate more Git LFS files (particularly, test data) in this repo, we are consuming the 300 GB monthly quota more and more quickly. One reason is that we currently run 18 jobs (6 upstream versions × 3 OSes), so even 100 MB of Git LFS files means a 1.8 GB hit towards that quota
  • This added, third-party action seems to be a response to others encountering the same issue. It appears the most developed, but is not 100% up to date; see some pending PRs https://github.com/nschloe/action-cached-lfs-checkout/pulls
  • We could mitigate by creating a fork or using someone else's.
• Configure Codecov to allow coverage to drop by 0.5% > 0.46% = 90 lines. This drop will (currently) always appear on PRs, because the snapshot and legacy reporting tests are only run on the main branch and a schedule trigger.
• Tidy a lot of nitpicks in docs.

How to review

Read the diff and note that the CI checks all pass.

PR checklist

• Continuous integration checks all ✅
• Add or expand tests; coverage checks both ✅
• Add, expand, or update documentation.
• Update doc/whatsnew.

[codecov]

Codecov Report

Attention: Patch coverage is 99.61832% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 75.5%. Comparing base (ad7d356) to head (7dc5385).

Additional details and impacted files

@@           Coverage Diff           @@
##            main    #162     +/-   ##
=======================================
- Coverage   76.0%   75.5%   -0.5%     
=======================================
  Files         94     101      +7     
  Lines       6428    6675    +247     
=======================================
+ Hits        4890    5046    +156     
- Misses      1538    1629     +91     

Files	Coverage Δ
message_ix_models/model/bare.py	100.0% <ø> (ø)
message_ix_models/model/build.py	100.0% <ø> (ø)
message_ix_models/model/disutility.py	100.0% <ø> (ø)
message_ix_models/model/emissions.py	98.0% <ø> (ø)
message_ix_models/model/macro.py	100.0% <ø> (ø)
message_ix_models/model/structure.py	100.0% <ø> (ø)
message_ix_models/project/shape/data.py	100.0% <100.0%> (ø)
message_ix_models/report/__init__.py	97.4% <100.0%> (+<0.1%)	⬆️
message_ix_models/report/compat.py	99.2% <ø> (ø)
message_ix_models/report/config.py	100.0% <ø> (ø)
... and 31 more

... and 4 files with indirect coverage changes

[glatterf42]

Thanks for these manifold improvements :)