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>

tests/test_agent_stream.py

@@ -0,0 +1,83 @@
+"""Live agent output: OpenRouter stream parsing and the SSE endpoint (status → delta → done)."""
+
+import asyncio
+import json
+
+import httpx
+import pytest
+from httpx import ASGITransport, AsyncClient
+
+from app.config import Settings
+from app.db import session
+from app.db.bootstrap import init_db
+from app.llm import agent
+from app.llm.agent import stream
+from app.llm.agent.providers import OpenRouterProvider
+from app.main import app
+
+MEM = "sqlite+aiosqlite:///:memory:"
+
+
+def _sse(*chunks: dict) -> bytes:
+    lines = [": OPENROUTER PROCESSING", ""]
+    lines += [f"data: {json.dumps(c, ensure_ascii=False)}" for c in chunks] + ["data: [DONE]", ""]
+    return "\n\n".join(lines).encode()
+
+
+async def test_openrouter_stream_deltas_and_tool_calls():
+    body = _sse(
+        {"choices": [{"delta": {"content": "Сейчас "}}]},
+        {
+            "choices": [
+                {
+                    "delta": {
+                        "tool_calls": [{"index": 0, "id": "c1", "function": {"name": "get_patient_", "arguments": ""}}]
+                    }
+                }
+            ]
+        },
+        {"choices": [{"delta": {"tool_calls": [{"index": 0, "function": {"name": "record", "arguments": "{}"}}]}}]},
+    )
+    client = httpx.AsyncClient(transport=httpx.MockTransport(lambda r: httpx.Response(200, content=body)))
+    provider = OpenRouterProvider(Settings(_env_file=None, openrouter_api_key="k"), client=client)
+    q: asyncio.Queue = asyncio.Queue()
+    token = stream.STREAM.set(q)
+    try:
+        turn = await provider.chat([{"role": "user", "content": "x"}], [])
+    finally:
+        stream.STREAM.reset(token)
+    assert [c.name for c in turn.tool_calls] == ["get_patient_record"]
+    events = [q.get_nowait() for _ in range(q.qsize())]
+    assert ("delta", "Сейчас ") in events and ("reset", None) in events
+
+
+@pytest.fixture
+async def client():
+    await session.configure(identity_url=MEM, clinical_url=MEM)
+    await init_db()
+    async with AsyncClient(transport=ASGITransport(app=app), base_url="http://t") as c:
+        yield c
+    await session.configure()
+
+
+async def test_sse_endpoint_streams_steps_then_final(client, monkeypatch):
+    async def fake(token, text, **kw):
+        stream.emit("status", stream.TOOL_STATUS["get_patient_record"])
+        stream.emit("delta", "Ответ ")
+        stream.emit("delta", "агента")
+        return agent.AgentAnswer(text="Ответ агента", sources=[], tools=["get_patient_record"], provider="fake")
+
+    monkeypatch.setattr(agent, "answer_chat", fake)
+    token = (await client.post("/api/v1/auth/demo/fresh", json={})).json()["token"]
+    kinds = []
+    async with client.stream(
+        "POST", f"/api/v1/dialog/{token}/events/stream", json={"type": "text", "text": "Можно ибупрофен?"}
+    ) as resp:
+        assert resp.headers["content-type"].startswith("text/event-stream")
+        async for line in resp.aiter_lines():
+            if line.startswith("event:"):
+                kinds.append(line[6:].strip())
+            if line.startswith("data:") and kinds and kinds[-1] == "done":
+                final = json.loads(line[5:])
+    assert kinds[0] == "status" and "delta" in kinds and kinds[-1] == "done"
+    assert any("Ответ агента" in r["text"] for r in final)