New test: Filtering tracks by length. Fixing a typo.

[hinderling]

This PR contains:

• a new test-case for the benchmark
  • I hereby confirm that NO LLM-based technology (such as github copilot) was used while writing this benchmark
• new dependencies in requirements.txt
  • The environment.yml file was updated using the command conda env export > environment.yml
• new generator-functions allowing to sample from other LLMs
• new samples (sample_....jsonl files)
• new benchmarking results (..._results.jsonl files)
• documentation update
• bug fixes

Related github issue (if relevant): ---

Short description:

• Added a new test-case for filtering tracks by length filter_tracks.
• Fixed a typo in a previous test apply_otsu_threshold_and_count_postiive_pixels -- > apply_otsu_threshold_and_count_positive_pixels.

How do you think will this influence the benchmark results?

• Adding more diversity to the tasks.
• Fixing the typo in the prompt might slightly improve the performance of some models.

Why do you think it makes sense to merge this PR?

• This is an often used task in my workflows. I usually use the trackpy library for this (as in the reference implementation I provided), and noticed that copilot seems to struggle to use this library or recreate the functionality. This is why I think it's an interesting test case. Generally I think processing tracking data is an integral part of bioimage analysis and not represented in this study yet, can try to add more examples if this is of interest!
• Fixing the typo increases legibility in result figures, and is more "fair" for model testing because there is no typo that could be confusing.

[haesleinhuepf]

Hi @hinderling ,

just a quick question regarding the notebook you sent. In the function definition there is a parameter "id_label='id'". In the docstring it says "id for each track (default: 'particle')". Isn't that a contradiction? Shouldn't it be "id" instead of particle?

Thanks!

Best,
Robert

[hinderling]

Oh no, thats a mistake, sorry! Should be fixed now, also changed the name of the test so other track filtering tasks (e.g. filter_tracks_by_value) can be added with less conflicting naming.