test(artwork): drop the decodeArtwork benchmark

It existed to capture the before/after baseline for the 128->100 thumbnail
change, which it has served. As an ongoing guard it is misleading: a 1000x1000
JPEG decode is ~97% of the 15ms it measures, so the two encoders it would be
reached for are ~2.6% of the signal. It reported a phantom 2.8% regression for
the NRGBA thumbnail change that isolating the scaler disproved.

The per-encoder benchmarks in blurhash/ and thumbhash/ cover the part that
actually changes.

core/artwork/processor_internal_bench_test.go

@@ -1,39 +0,0 @@
-package artwork
-
-import (
-	"bytes"
-	"context"
-	"image"
-	"image/color"
-	"image/jpeg"
-	"testing"
-)
-
-func benchJPEG(size int) []byte {
-	img := image.NewRGBA(image.Rect(0, 0, size, size))
-	for y := range size {
-		for x := range size {
-			img.SetRGBA(x, y, color.RGBA{
-				R: uint8(255 * x / size), G: uint8(255 * y / size),
-				B: uint8((x + y) * 255 / (2 * size)), A: 255,
-			})
-		}
-	}
-	var buf bytes.Buffer
-	if err := jpeg.Encode(&buf, img, nil); err != nil {
-		panic(err)
-	}
-	return buf.Bytes()
-}
-
-// BenchmarkDecodeArtwork measures the whole per-image cost the artwork worker pays.
-func BenchmarkDecodeArtwork(b *testing.B) {
-	data := benchJPEG(1000)
-	ctx := context.Background()
-	b.ReportAllocs()
-	for b.Loop() {
-		if _, err := decodeArtwork(ctx, "bench", data); err != nil {
-			b.Fatal(err)
-		}
-	}
-}