Feat/standardize benchmarks

[miagarvey]

Standardize fa, pheno, drug, study params benchmarks

⚙️ Release Notes

• Create shared_utils.py with shared evaluation functions (exact_match, semantic_similarity, category_equal, compute_weighted_score); refactor FA, Drug, and Study Parameters benchmarks to use them.
• Add optional field_weights, dependency validation (Drug: direction/association checks; Study Parameters: statistical consistency), and standardized allele handling (semantic_similarity) across all benchmarks.
• Standardize Pheno benchmark: return detailed dict (0-1 scale) instead of single float (0-100), use field-specific evaluators from shared_utils, add field_weights, and update scaffold to accept ground truth directly (backward compatible with PMCID fallback).

By submitting this pull request, you agree to follow our Coding Guidelines:

• I agree to follow the Coding Guidelines.