fix(evals)!: remove positional arguments in favor of kwargs

[ehutt]

Enforces kwargs everywhere except the .evaluate and .async_evaluate methods on Evaluators which only require one argument and have one optional argument.

Also removes the evaluator registry which was not very useful and keeps implementation simpler.

---

Note

Enforces keyword-only parameters across the evals public API and removes the evaluator registry/list_evaluators, updating code, tests, and tutorials accordingly.

• API changes (keyword-only):
  • Enforce kwargs in Evaluator.__init__, LLMEvaluator.__init__, ClassificationEvaluator.__init__, create_evaluator, create_classifier, bind_evaluator, evaluate_dataframe, async_evaluate_dataframe, utils.remap_eval_input, utils.to_annotation_dataframe.
  • Update internal and caller sites to pass named args (e.g., evaluate(eval_input=..., input_mapping=...)).
• Removals:
  • Delete evaluator registry (_registry) and list_evaluators; drop exports from __init__.py and evaluators.__all__.
• Tests & tutorials:
  • Refactor tests and notebooks to use the new keyword-only API; remove registry tests and adjust examples (bind_evaluator, evaluate_dataframe, to_annotation_dataframe, etc.).

^{Written by Cursor Bugbot for commit c9af39b. This will update automatically on new commits. Configure here.}

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[cursor]

Bug: Nonexistent Python Version in Notebook Metadata

Invalid Python version "3.9.23" in Jupyter notebook metadata. Python version 3.9.23 doesn't exist - the latest Python 3.9.x version is 3.9.21. This appears to be accidentally committed notebook metadata that could cause issues if tools try to validate or use this version number.