[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

tests/test_live_runtime.py

@@ -97,7 +97,7 @@ def make_runtime(texts: list[str], context: ConversationContext | None = None) -
     reporter = RecordingReporter()
     event_bus = PipelineEventBus()
     runtime = VoiceAssistantPipeline(
-        config=AppConfig(llm_api_key="secret", speech_provider="cloud", post_playback_drain_ms=0),
+        config=AppConfig(llm_api_key="secret", speech_provider="cloud"),
         transport=transport,
         wakeword=KeywordWakeWordProvider(),
         vad_recorder=VadRecorder(EnergyVadProvider(), min_duration_ms=40, end_silence_ms=40),
@@ -119,6 +119,7 @@ class LiveRuntimeTests(unittest.TestCase):
         summary = runtime.run(max_turns=2)
 
         self.assertEqual(summary.completed_turns, 2)
+        self.assertEqual(runtime.config.post_playback_drain_ms, 0)
         self.assertEqual(len(stt.calls), 2)
         self.assertEqual(len(llm.calls), 2)
         self.assertEqual(len(transport.played_segments), 4)