Function linear_score 

pub fn linear_score(
    vec_dist: f32,
    min_vec: f32,
    max_vec: f32,
    bm25: f32,
    min_bm25: f32,
    max_bm25: f32,
    bm25_weight: f32,
) -> f32

Apply linear fusion over normalized (vec_dist, bm25_score) pairs.

Both inputs are normalized to [0,1] across the candidate set before fusion. Returns value in [0,1] where 0 = best (lower = better convention).