Don't recreate the speech encoder if we don't have to

webrtc/modules/audio_coding/codecs/cng/audio_encoder_cng.h

Index: webrtc/modules/audio_coding/codecs/cng/audio_encoder_cng.h
diff --git a/webrtc/modules/audio_coding/codecs/cng/audio_encoder_cng.h b/webrtc/modules/audio_coding/codecs/cng/audio_encoder_cng.h
index 246e9519da486a1e7a35d26d9d59042cb1461229..a895e69de44b4cb97819d3f944875b6d9aae948d 100644
--- a/webrtc/modules/audio_coding/codecs/cng/audio_encoder_cng.h
+++ b/webrtc/modules/audio_coding/codecs/cng/audio_encoder_cng.h
@@ -63,6 +63,8 @@ class AudioEncoderCng final : public AudioEncoder {
   void SetMaxPlaybackRate(int frequency_hz) override;
   void SetProjectedPacketLossRate(double fraction) override;
   void SetTargetBitrate(int target_bps) override;
+  rtc::ArrayView<std::unique_ptr<AudioEncoder>> ReclaimContainedEncoders()
+      override;
 
  private:
   EncodedInfo EncodePassive(size_t frames_to_encode,