feat: improve voice studio alpha recovery flow

backend/app/schemas/voice_session_schemas.py

@@ -101,6 +101,7 @@ class VoiceSessionSummaryResponse(BaseModel):
     latest_detected_intent: str | None = None
     latest_assistant_audio_ready: bool = False
     last_turn_status: str | None = None
+    transcription_mode_hint: str | None = None
     can_continue: bool = False
     can_finalize: bool = False
     last_error: str | None = None