[唤醒应答与快速端点]：完成唤醒后我在播报和本地VAD端点优化，包含缓冲清理、配置项和测试覆盖

src/owner_voice_pet/transport.py

@@ -103,6 +103,11 @@ class MemoryAudioTransport:
         self.played_segments.append(segment)
         return PlaybackResult(True, segment.duration_ms)
 
+    def flush_input(self) -> int:
+        count = len(self._frames)
+        self._frames.clear()
+        return count
+
     def stop(self) -> None:
         self.started = False
 
@@ -290,6 +295,16 @@ class SoundDeviceAudioTransport:
                 break
         return None
 
+    def flush_input(self) -> int:
+        count = 0
+        while not self._queue.empty():
+            try:
+                self._queue.get_nowait()
+                count += 1
+            except queue.Empty:
+                break
+        return count
+
     def health(self) -> TransportHealth:
         if self._sd is None:
             return TransportHealth(False, False, "sounddevice unavailable")