feat(wing): score a wing by its best page, not summed hits; add wingbacktest command (#205 S1)

A first backtest showed BM25 summed over a wing's top hits favours big wings
(0/4 agreement). Best-page scoring is volume-robust; pinned by a test that
failed before the change.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

ingestion/internal/wing/bm25scorer.go

@@ -11,7 +11,6 @@ import (
 const (
 	maxQueryTerms = 12
 	minTermLen    = 5
-	hitsPerWing   = 3
 )
 
 // BM25Scorer scores a note against each wing by running the brain's own BM25
@@ -38,24 +37,18 @@ func (b *BM25Scorer) Score(text string) map[string]float64 {
 		return out
 	}
 	for _, w := range b.wings {
-		hits, err := search.Query(b.brainDir, search.QueryOptions{Query: query, Wing: w, Limit: hitsPerWing + 1})
+		hits, err := search.Query(b.brainDir, search.QueryOptions{Query: query, Wing: w, Limit: 2})
 		if err != nil {
 			continue
 		}
-		var sum float64
-		n := 0
 		for _, h := range hits {
 			if b.exclude != "" && h.Path == b.exclude {
 				continue
 			}
-			if n == hitsPerWing {
-				break
-			}
-			sum += float64(h.Score)
-			n++
-		}
-		if sum > 0 {
-			out[w] = sum
+			// Hits come best-first: a wing is scored by its single best
+			// page, so a big wing cannot win on volume alone.
+			out[w] = float64(h.Score)
+			break
 		}
 	}
 	return out