[实时转写]：完成录音期间转写显示，包含partial事件、本地streaming STT和终端实时反馈

src/owner_voice_pet/protocols.py

@@ -68,6 +68,19 @@ class SttProvider(Protocol):
         ...
 
 
+class RealtimeTranscriptSession(Protocol):
+    def accept_frame(self, frame: AudioFrame) -> Transcript | None:
+        ...
+
+    def finish(self) -> Transcript | None:
+        ...
+
+
+class RealtimeSttProvider(SttProvider, Protocol):
+    def start_stream(self) -> RealtimeTranscriptSession:
+        ...
+
+
 class LlmProvider(Protocol):
     def stream_reply(self, messages: Sequence[Message]) -> Iterable[ReplyDelta]:
         ...