Function mig_select 

pub fn mig_select(
    scores: &[SaliencyScore],
    chunks: &[ContentChunk],
    top_k: usize,
    lambda: f64,
) -> Vec<usize>

Select top-k chunks using MIG: Marginal Information Gain.

Greedily selects chunks that maximize relevance while minimizing redundancy with already-selected chunks.

lambda: trade-off between relevance and diversity (0.0 = pure relevance, 1.0 = pure diversity). Default: 0.6.