feat:medcalculator toolkit

[ZackZikaiXiao]

PR Description: Modify and Update Medicine Toolkit (verifier)

Related Issues:

Resolves #2178.

Changes Made:

• Rewrote medcalc_bench tools:

  • Included all tool functions from medcalc_bench with enhanced implementation.
  • Added standardized docstrings to clearly describe the functionality, parameters, and return values of each tool.
  • Provided test cases and practical examples for validating each tool’s correctness and usability.
  • Fixed existing bugs and errors identified within the original implementation.
  • Improved the directory structure of medcalc_bench, distinctly organizing tool modules and foundational utility modules.

• Created medcalc_toolkit.py:

  • Developed a centralized interface for invoking all rewritten tools.
  • Ensured each function includes clear, standardized docstrings.

• Added a Toolkit Example (tool_example.py):

  • Included detailed usage examples demonstrating how to effectively utilize the medcalc_toolkit.

Testing Details:

• Conducted testing using one data sample per tool, totaling 44 tools tested.
• Noted 4-5 tools lacking official test data from medcalc_bench; thus, these tools have not been verified.
• Identified discrepancies between gold labels and inference results for approximately 3 tools, requiring further review.

Motivation:

This update directly addresses the feature request raised in #2178, aiming to improve maintainability, usability, and reliability within the Loong project's medical calculation components.

How to Test:

1. Verify availability and documentation quality of tools within medcalc_toolkit.py.
2. Run provided test cases to ensure functionality and accuracy.
3. Review tool_example.py to validate practical usage scenarios.
4. Pay special attention to tools identified with test data issues or discrepancies noted above.

Additional Notes:

• PR addresses multiple issues highlighted during initial assessments, particularly related to accuracy and organizational clarity.
• Careful review of untested tools or those showing result mismatches is advised during code review.

[lightaime]

Thanks @ZackZikaiXiao @13204942 for contributing this pull request! Please also make sure we follow their license during the development: https://github.com/ncbi-nlp/MedCalc-Bench/blob/main/CODE-LICENSE.

[zjrwtx]

thanks @ZackZikaiXiao @13204942 ,please fix the pre-commit error