test(thumbhash): drop the shipped-vs-reference benchmark

The two-pass rewrite's advantage over the naive port is already recorded in its
commit message; carrying the benchmark to re-derive it has no ongoing use.

reference_test.go stays: besides the benchmark baseline it holds the
differential oracle the golden-vector and randomised specs assert against.

core/artwork/thumbhash/thumbhash_bench_test.go

@@ -1,36 +0,0 @@
-package thumbhash_test
-
-import (
-	"testing"
-
-	"github.com/navidrome/navidrome/core/artwork/thumbhash"
-	"github.com/navidrome/navidrome/tests"
-)
-
-// BenchmarkEncodeVsReference pins the two-pass separable encoder against the naive reference port
-// it replaced. referenceEncode is test-only, so this cannot live beside the cross-encoder
-// benchmark in core/artwork.
-func BenchmarkEncodeVsReference(b *testing.B) {
-	const size = 100
-	img := tests.GradientImage(size)
-	pix := make([]byte, 0, size*size*4)
-	for y := range size {
-		pix = append(pix, img.Pix[y*img.Stride:y*img.Stride+size*4]...)
-	}
-
-	b.Run("shipped", func(b *testing.B) {
-		b.ReportAllocs()
-		for b.Loop() {
-			if _, err := thumbhash.Encode(img); err != nil {
-				b.Fatal(err)
-			}
-		}
-	})
-
-	b.Run("reference", func(b *testing.B) {
-		b.ReportAllocs()
-		for b.Loop() {
-			_ = referenceEncode(size, size, pix)
-		}
-	})
-}