feat: Token-Level-LLM-Streaming über WebSocket (#4 Ausbau)

- LLMProvider.stream: Basis-Default (Fallback über complete) + SSE-Streaming
  fuer OpenRouter und lokalen OpenAI-kompatiblen Provider; gemeinsamer Parser sse_delta
- Orchestrator.chat_stream: LLM-Token live via on_token-Callback, danach
  Spoken-Adapter/Normalizer/TTS/Output; Fallback fuer Provider ohne stream
- WS /ws/chat: opt-in {"stream":true} -> ack -> token* -> semantic -> audio -> done
- Tests: 43 gruen (+5: SSE-Parsing, Default-Fallback, WS-Token-Flow)
- Doku aktualisiert; .gitignore: *.wav (generierte Audio-Ausgaben)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

app/providers/llm/openrouter.py

@@ -1,6 +1,8 @@
+from collections.abc import AsyncIterator
+
 import httpx
 
-from app.providers.llm.base import LLMProvider
+from app.providers.llm.base import LLMProvider, sse_delta
 
 
 SYSTEM_PROMPT = """
@@ -49,12 +51,7 @@ class OpenRouterLLMProvider(LLMProvider):
         self.api_key = (api_key or "").strip()
         self.model = (model or "").strip()
 
-    async def complete(
-        self,
-        text: str,
-        history: list[dict] | None = None,
-        session_id: str | None = None,
-    ) -> str:
+    def _build_messages(self, text: str, history: list[dict] | None) -> list[dict]:
         if not self.api_key:
             raise ValueError("OPENROUTER_API_KEY is empty")
         if not self.model:
@@ -66,10 +63,17 @@ class OpenRouterLLMProvider(LLMProvider):
         if history:
             messages.extend(history)
         messages.append({"role": "user", "content": text.strip()})
+        return messages
 
+    async def complete(
+        self,
+        text: str,
+        history: list[dict] | None = None,
+        session_id: str | None = None,
+    ) -> str:
         payload = {
             "model": self.model,
-            "messages": messages,
+            "messages": self._build_messages(text, history),
         }
 
         timeout = httpx.Timeout(connect=10.0, read=120.0, write=30.0, pool=10.0)
@@ -106,3 +110,42 @@ class OpenRouterLLMProvider(LLMProvider):
 
         return str(content).strip()
 
+    async def stream(
+        self,
+        text: str,
+        history: list[dict] | None = None,
+        session_id: str | None = None,
+    ) -> AsyncIterator[str]:
+        payload = {
+            "model": self.model,
+            "messages": self._build_messages(text, history),
+            "stream": True,
+        }
+        timeout = httpx.Timeout(connect=10.0, read=120.0, write=30.0, pool=10.0)
+
+        async with httpx.AsyncClient(timeout=timeout) as client:
+            try:
+                async with client.stream(
+                    "POST",
+                    "https://openrouter.ai/api/v1/chat/completions",
+                    headers={
+                        "Authorization": f"Bearer {self.api_key}",
+                        "Content-Type": "application/json",
+                    },
+                    json=payload,
+                ) as response:
+                    if response.status_code >= 400:
+                        body = await response.aread()
+                        raise RuntimeError(
+                            f"OpenRouter LLM error {response.status_code}: "
+                            f"{body.decode(errors='replace')}"
+                        )
+                    async for line in response.aiter_lines():
+                        delta = sse_delta(line)
+                        if delta:
+                            yield delta
+            except httpx.TimeoutException as exc:
+                raise RuntimeError("OpenRouter LLM timeout") from exc
+            except httpx.HTTPError as exc:
+                raise RuntimeError(f"OpenRouter LLM transport error: {exc}") from exc
+