fix: Add namaa MrTydi reranking dataset

[omarelshehy]

Why this dataset:

1 - Add to the reranking tasks exclusively for arabic
2 - Utilize the test dataset for MrTydi with generated and human-evaluated negatives.

Checklist

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

Adding datasets checklist

Reason for dataset addition: ...

• I have run the following models on the task (adding the results to the pr). These can be run using the mteb -m {model_name} -t {task_name} command.
  • cross-encoder/ms-marco-MiniLM-L-12-v2
  • cross-encoder/stsb-TinyBERT-L-4
• I have checked that the performance is neither trivial (both models gain close to perfect scores) nor random (both models gain close to random scores).
• If the dataset is too big (e.g. >2048 examples), considering using self.stratified_subsampling() under dataset_transform()
• I have filled out the metadata object in the dataset file (find documentation on it here).
• Run tests locally to make sure nothing is broken using make test.
• Run the formatter to format the code using make lint.

[Samoed]

Did your dataset add new data to the original MrTidy? The original MrTidy is already included in MTEB. Also, could you provide results for these tasks to ensure it's working correctly? It seems like the data is being loaded in a different format than expected

[omarelshehy]

I might be mistaken, but the Mrtydi dataset was included there for retrieval and not reranking. We basically took the test dataset from MrTydi and added 4-5 negatives to each query and positive (which the original doesn't have). For the formatting I relied on similar Reranking dataset structures. Here is also the results of the two models in the PR description
NamaaMrTydiReranking_ms-marco-MiniLM.json
NamaaMrTydiReranking_stsb_TinyBERT.json

[Samoed]

Ah, yes. You are right

[KennethEnevoldsen]

The metadata seems to be lacking a bit, I have suggested some updates.