feat: live agent answers (SSE streaming) + «Чат» tab on phones

- POST /dialog/{token}/events/stream: Server-Sent Events with the agent's steps («Смотрю вашу карту ЕМИАС…»,
  «Ищу в клинических рекомендациях…»), the answer as the model writes it (OpenRouter stream=true), then the
  final replies (same as /events, incl. the forbidden-output filter and sources). X-Accel-Buffering: no.
- Web: free-text messages use the stream — a live bubble with steps, typing dots and a cursor; falls back to
  the plain request if the stream breaks.
- Phones: the bottom tab bar gets «Чат» (it had no way back to the chat).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

app/llm/agent/providers.py

@@ -16,6 +16,7 @@ import httpx
 from pydantic import BaseModel, Field
 
 from app.config import Settings
+from app.llm.agent import stream
 
 log = logging.getLogger(__name__)
 
@@ -53,6 +54,8 @@ class OpenRouterProvider:
         self._client = client
 
     async def chat(self, messages: list[dict[str, Any]], tools: list[dict[str, Any]]) -> AssistantTurn:
+        if stream.active():
+            return await self._chat_stream(messages, tools)
         body: dict[str, Any] = {"model": self._model, "messages": messages, "tools": tools, "temperature": 0.1}
         body["max_tokens"] = 900
         if not any(m["role"] == "tool" for m in messages):
@@ -174,6 +177,78 @@ class GigaChatToolProvider:
         return AssistantTurn(content=msg.content or None, tool_calls=calls)
 
 
+# ------------------------------------------------------------------------------------------------ streaming
+
+
+async def _iter_sse(resp: httpx.Response):
+    async for line in resp.aiter_lines():
+        if not line.startswith("data:"):
+            continue  # comments («: OPENROUTER PROCESSING») and blank lines
+        data = line[5:].strip()
+        if data == "[DONE]":
+            return
+        try:
+            yield json.loads(data)
+        except json.JSONDecodeError:
+            continue
+
+
+async def _openrouter_stream(self: "OpenRouterProvider", messages: list[dict[str, Any]], tools: list[dict[str, Any]]):
+    """Same request with `stream: true`: content deltas go to the live channel, tool calls are assembled."""
+    body: dict[str, Any] = {
+        "model": self._model,
+        "messages": messages,
+        "tools": tools,
+        "temperature": 0.1,
+        "max_tokens": 900,
+        "stream": True,
+    }
+    if not any(m["role"] == "tool" for m in messages):
+        body["tool_choice"] = "required"
+    headers = {"Authorization": f"Bearer {self._key}", "X-Title": "PTSRaiAgent prototype (synthetic data)"}
+    content: list[str] = []
+    calls: dict[int, dict[str, Any]] = {}
+    try:
+        client = self._client or httpx.AsyncClient(timeout=self._timeout, proxy=self._proxy)
+        try:
+            async with client.stream("POST", self._url, json=body, headers=headers, timeout=self._timeout) as resp:
+                if resp.status_code != 200:
+                    await resp.aread()
+                    raise ProviderError(f"http {resp.status_code}")
+                async for chunk in _iter_sse(resp):
+                    choice = (chunk.get("choices") or [{}])[0]
+                    delta = choice.get("delta") or {}
+                    if delta.get("content"):
+                        content.append(delta["content"])
+                        stream.emit("delta", delta["content"])
+                    for tc in delta.get("tool_calls") or []:
+                        slot = calls.setdefault(tc.get("index", 0), {"id": "", "name": "", "args": ""})
+                        slot["id"] = tc.get("id") or slot["id"]
+                        fn = tc.get("function") or {}
+                        slot["name"] += fn.get("name") or ""
+                        slot["args"] += fn.get("arguments") or ""
+        finally:
+            if self._client is None:
+                await client.aclose()
+    except httpx.HTTPError as exc:
+        raise ProviderError(type(exc).__name__) from exc
+    out = []
+    for i in sorted(calls):
+        c = calls[i]
+        try:
+            args = json.loads(c["args"] or "{}")
+        except json.JSONDecodeError:
+            args = {}
+        out.append(ToolCall(id=c["id"] or c["name"] or f"call{i}", name=c["name"], arguments=args))
+    if out and content:
+        stream.emit("reset", None)  # a preface before tool calls is not the answer
+    log.info("openrouter stream ok calls=%d chars=%d", len(out), sum(map(len, content)))
+    return AssistantTurn(content="".join(content) or None, tool_calls=out)
+
+
+OpenRouterProvider._chat_stream = _openrouter_stream  # type: ignore[attr-defined]
+
+
 # ------------------------------------------------------------------------------------------------ offline mock
 
 _DRUG_WORDS = re.compile(