refactor: Use ElevenLabs directly for both STT and TTS

- STT: elevenlabs.STT() (Scribe) instead of Whisper via LiteLLM
- TTS: elevenlabs.TTS() (already direct)
- LLM: still routed through LiteLLM/OpenRouter
- No extra API accounts needed — only ElevenLabs + OpenRouter

CF-1147

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

agent.py

@@ -32,11 +32,7 @@ async def entrypoint(ctx: JobContext):
     voice_id = os.environ.get("ELEVENLABS_VOICE_ID", "21m00Tcm4TlvDq8ikWAM")
 
     session = AgentSession(
-        stt=lk_openai.STT(
-            base_url=LITELLM_URL,
-            api_key=LITELLM_KEY,
-            model="whisper",
-        ),
+        stt=elevenlabs.STT(),
         llm=lk_openai.LLM(
             base_url=LITELLM_URL,
             api_key=LITELLM_KEY,