Implement safeguard against problematic meta keys

[danielhuppmann]

Just ran into an issue (with @pmussak) in the new ECEMF app and ixmp4 database instance - I had added meta-indicators named "version" to the database, which was "overwriting" the version-attribute in the run-selection component of the app.

To avoid such issues, this PR makes several terms illegal as meta-indicator-keys (at creation). I briefly thought about adding this to the meta-model (see link below), but I was worried that if a meta indicator with such a name was already created in a database with a current ixmp4 version, there might be issues with then loading them after upgrading ixmp4.

ixmp4/ixmp4/data/db/meta/model.py

Line 64 in 70366d4

def __init__(self, *args, **kwargs) -> None:

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.9%. Comparing base (70366d4) to head (4be6149).
Report is 3 commits behind head on main.

Additional details and impacted files

@@          Coverage Diff          @@
##            main    #131   +/-   ##
=====================================
  Coverage   85.8%   85.9%           
=====================================
  Files        227     227           
  Lines       7564    7583   +19     
=====================================
+ Hits        6494    6514   +20     
+ Misses      1070    1069    -1     

Files with missing lines	Coverage Δ
ixmp4/data/db/meta/repository.py	96.0% <100.0%> (+0.1%)	⬆️

... and 9 files with indirect coverage changes

---- 🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[meksor]

If I am not mistaken, a check in RunMetaEntryRepository.bulk_upsert is missing!

[meksor]

Suggested change

	def test_illegal_key(self, platform: ixmp4.Platform):
	run = platform.runs.create("Model", "Scenario")
	with pytest.raises(InvalidRunMeta):
	platform.backend.meta.create(run.id, "version", "foo")
	def test_illegal_key(self, platform: ixmp4.Platform):
	run = platform.runs.create("Model", "Scenario")
	for illegal_key in {"model", "scenario", "id", "version", "is_default"}:
	with pytest.raises(InvalidRunMeta):
	platform.backend.meta.create(run.id, illegal_key, "foo")

Might be good to test all the keys from a distinct set here so we have to think about what we broke when removing keys from the actual set.

[danielhuppmann]

Added a loop for testing all illegal keys.

[meksor]

Also not quite sure what exactly the problem is, i think the api currently does not allow for filtering my meta indicators... So is this check here now to account for something going on in the frontend implementation of the data explorer?

[danielhuppmann]

Thanks @meksor, added the check also to bulk_upsert() (with a test).

For clarification, the aim of this PR is to prevent users from setting meta indicators that could be confusing, irrespective of whether it's confusing for the front-end API or for other users.

[meksor]

I see