[低延迟端点]：完成首句保留和快速结束修正，包含ACK缓冲策略、批量读帧和主说话人端点回归测试

src/owner_voice_pet/transport.py

@@ -223,9 +223,15 @@ class SoundDeviceAudioTransport:
             return []
         timeout_s = max(0, timeout_ms) / 1000
         try:
-            return [self._queue.get(timeout=timeout_s)]
+            frames = [self._queue.get(timeout=timeout_s)]
         except queue.Empty:
             return []
+        while True:
+            try:
+                frames.append(self._queue.get_nowait())
+            except queue.Empty:
+                break
+        return frames
 
     def play_pcm(self, segment: AudioSegment, interrupt: bool = False) -> PlaybackResult:
         if self._sd is None: