BFCL metric bug fix

[aman-servicenow]

Fix: Scalar param and required param check bug

📌 Description

🔗 Related Issue(s)

🛠️ Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality including new tasks)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Refactor / Code cleanup
• Maintenance / Chore / Task
• Other (please describe):

✅ How Has This Been Tested?

• Unit tests
• Integration tests
• Manual testing

Test Results / Screenshots (if applicable):

📸 Screenshots / Demos

📋 Checklist

• Code follows project style guidelines
• Tests have been added/updated (if applicable)
• Documentation has been updated (if applicable)
• Linked relevant issue(s)
• Self-reviewed my code

🙌 Additional Notes

Fixes issues in BFCL metric:

1. Scalar parameter values are never validated

The _compare_tool_call method only checks values for dict and list typed parameters. For scalar types (string, integer, float, boolean), it only verifies the parameter exists — never comparing the predicted value against the ground truth. Since the vast majority of BFCL parameters are scalars, a model outputting base=999 when the answer is base=10 would still score as correct. In the standard BFCL evaluator, simple_function_checker has a catch-all value not in possible_answer[param] that correctly rejects wrong values for all types. Note that simple_function_checker is the core per-call validator used by all categories — multiple_function_checker and parallel_function_checker_no_order both delegate to it - so this affects simple, multiple, parallel, and parallel_multiple equally.

2. Only required parameters are checked
   The metric loop iterates only over required_params from the tool schema. Optional parameters with expected values in the ground truth (e.g., "unit": ["units", ""]) are never validated. In contrast, standard BFCL's simple_function_checker iterates over all model_params.items() and also checks for missing optional parameters that are not marked as optional in the ground truth. Note that simple_function_checker is the core per-call validator used by all categories - multiple_function_checker and parallel_function_checker_no_order both delegate to it - so this affects simple, multiple, parallel, and parallel_multiple equally.