fix(tts): revert to multilingual_v2 for better quality, keep speed 1.15x

flash_v2_5 had audible compression artifacts. multilingual_v2 has higher
fidelity while speed=1.15 via VoiceSettings still gives snappier delivery.

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

voice.py

@@ -588,7 +588,7 @@ class VoiceSession:
                 llm=lk_openai.LLM(base_url=LITELLM_URL, api_key=LITELLM_KEY, model=self.model),
                 tts=elevenlabs.TTS(
                     voice_id=voice_id,
-                    model="eleven_flash_v2_5",
+                    model="eleven_multilingual_v2",
                     language="de",
                     api_key=ELEVENLABS_KEY,
                     voice_settings=elevenlabs.VoiceSettings(