Add filtered ffqa dataset

[marukaz]

Add WikiQA-Free_Form_QA dataset, a long context QA dataset.
This dataset is formatted to be used in chat format, but I want to experiment with 8k context LLM, so I implemented preprocess_steps accordingly.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 88.46%. Comparing base (7b50122) to head (6ade2f1).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #593   +/-   ##
=======================================
  Coverage   88.46%   88.46%           
=======================================
  Files          87       87           
  Lines        7779     7779           
=======================================
  Hits         6882     6882           
  Misses        897      897           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[yoavkatz]

@elronbandel - why does the catalog preparation this fail with:

[yoavkatz]

The failure due to the fact that the this is merge from a fork. We will handle this.

[elronbandel]

Issue is now fixed @marukaz . Notice i rebased your branch to re-run the tests.

[elronbandel]

Really cool contribution! This dataset will be very hard to incorpirate to unitxt without the creativity you showed with the ExcuteExpression! so thank you! If you can change the things i asked it will be great. let me know when your ready and ill merge.

[marukaz]

@elronbandel Thank you for your review! I updated code.

[marukaz]

catalog preparation test is failing but when I tried python -m unittest discover -s tests/catalog -p "test_*.py" in my local it succeeded. I would like to rerun the test but I don't have the permission.