Function run_scoring 

pub fn run_scoring(
    features: &[Feature],
    scoring_cfg: &ScoringConfig,
    features_cfg: &FeaturesConfig,
    polarity: Polarity,
) -> Vec<ScoredFeature>

Stage 3: Score isotope features using the averagine model.

Applies three AND-ed retention filters from FeaturesConfig:

• min_isotope_score (Bhattacharyya vs averagine)
• min_cosine_score (mean chromatographic cosine of isotope hills)
• min_combined_score (= isotope × cosine)

All thresholds at 0.0 keep every scored feature.