fix: Fix test for empty descriptive tasks

[Samoed]

Checklist

• Run tests locally to make sure nothing is broken using make test.
• Run the formatter to format the code using make lint.

It seems that when creating the test, I missed recreating the array in the test case, which caused the test to not work correctly. Fixes #1412

[KennethEnevoldsen]

This of course causes the tests to fail, if possible it would be great to run calculate descriptive stats for all datasets (maybe with a few exceptions).

[Samoed]

Oh, I missed that test not passing. I'm preparing an update to descriptive stats, and after that I think I can try to calculate on some datasets

[KennethEnevoldsen]

Moving this to v2.0.0, but we could move it back if it makes things easier

[Samoed]

@KennethEnevoldsen Tests are fixed. I think this is ready to merge into main for now. If anyone wants to add new tasks, this test should be helpful. Future updates to descriptive stats could be planned for version 2.0.0.

[KennethEnevoldsen]

Shouldn't the test fail? GIven that most datasets don't have descriptive stats?

[Samoed]

I've added current tasks without metadata to exceptions like in test_disallow_trust_remote_code_in_new_datasets

[KennethEnevoldsen]

Hmm, we might be misunderstanding each other. A task like "TbilisiCityHallBitextMining" no longer has descriptive statistics, but did have it beforehand.

[Samoed]

I updated the test. Now, if a task is in the exceptions list (tasks currently without descriptive stats) and has metadata, it will raise an error. Otherwise, descriptive stats are required.