fix(scraper): add Brotli decompression to HTTP client

novelfire.net responds with Content-Encoding: br when the scraper
advertises 'gzip, deflate, br'. The client only handled gzip, so
Brotli-compressed bytes were fed raw into the HTML parser producing
garbage — empty titles, zero chapters, and selector failures.

Added github.com/andybalholm/brotli and wired it into GetContent
alongside the existing gzip path.

scraper/internal/browser/http.go

@@ -10,6 +10,8 @@ import (
 	"os"
 	"strings"
 	"time"
+
+	"github.com/andybalholm/brotli"
 )
 
 type httpClient struct {
@@ -106,16 +108,17 @@ func (c *httpClient) GetContent(ctx context.Context, req ContentRequest) (string
 	// net/http decompresses gzip automatically only when it sets the header
 	// itself; since we set Accept-Encoding explicitly we must do it ourselves.
 	body := resp.Body
-	if strings.EqualFold(resp.Header.Get("Content-Encoding"), "gzip") {
+	switch strings.ToLower(resp.Header.Get("Content-Encoding")) {
+	case "gzip":
 		gr, gzErr := gzip.NewReader(resp.Body)
 		if gzErr != nil {
 			return "", fmt.Errorf("http: gzip reader: %w", gzErr)
 		}
 		defer gr.Close()
 		body = gr
+	case "br":
+		body = io.NopCloser(brotli.NewReader(resp.Body))
 	}
-	// br (Brotli) decompression requires an external package; skip for now —
-	// the server will fall back to gzip or plain text for unknown encodings.
 
 	raw, err := io.ReadAll(body)
 	if err != nil {