Remove 'batch_size' from fw_kwargs in WhisperTranscriber to streamline transcription options.

app.py

@@ -653,6 +653,7 @@ class WhisperTranscriber:
                     **fw_kwargs,
                 )
             else:
+                fw_kwargs.pop("batch_size", None)
                 segments_iter, info = fw_model.transcribe(
                     audio,
                     language=language,