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): closes #120

Short description:

• All of these changes perform the same function - moving the numpy array assertion to the check block rather than the function block. This brings our test-cases in line with each other, and addresses a substantial part of Check why rather basic test-cases are failing #76 by fixing basic test-case checks.

How do you think will this influence the benchmark results?
This changes the following test-case pass-rates on my model evaluation of the existing prompts:

• mask_images pass-rate from 31 / 230 -> 176 / 230.
• sum_images pass-rate from 11 / 230 -> 67 / 230.
• mean_squared_error pass-rate from 21 / 230 -> 167 / 230.
• rgb_to_grey_image_transform from 6 / 230 -> 64 / 230

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

• This PR unifies these test-cases both with their current prompts and other test-cases (which already assume this sort of function IO)