✨ feat: WP-05 模型网关核心——逻辑模型、流式适配与 Run 真实执行

- 迁移 7+8：logical_model（能力目录）、provider_endpoint（密钥只存环境变量名，永不落库）、turn、model_call（每 attempt 单独持久化：usage/TTFT/错误，Provider 未返回保持 null 不估算——ADR-003）；message.run_id 关联
- OpenAI-compatible 流式适配器：httpx SSE 解析、标准化 StreamChunk（delta/usage/done）、transport 可注入（测试与故障注入）；密钥经环境变量读取不进日志
- Run 执行器：后台执行完整生命周期——running → content.block.started → 真实 content.block.delta*（流式产生，delta_index 单调）→ content.block.completed → message.completed → 终态；assistant 消息为权威快照；ModelCall 记录 usage 与 TTFT；无可用端点时 Run 显式失败并留 error 事件（fail-closed）
- post_message：响应提交后由框架 BackgroundTasks 执行 Run（显式预提交保证 Run 落库）；SSE 改为增量轮询（重放 + 等待新事件至终态，30s 上限）
- 测试 54 项全绿（+2）：端到端 mock 流验证 ModelCall usage/TTFT/assistant 权威消息、无端点 Run failed + content.block.failed 事件

backend/tests/conftest.py

@@ -46,6 +46,28 @@ async def client(db):
                 raise
 
     app.dependency_overrides[db_session.get_db] = override
+    # 后台 Run 任务使用测试库会话工厂与 mock 流（生产为独立工厂与真实传输）
+    app.state.session_factory = db
+    app.state.gw_transport = _mock_openai_stream()
+    import os as _os
+
+    _os.environ.setdefault("HJ_MOCK_KEY", "test-key")
+    from app.models import LogicalModel, ProviderEndpoint
+
+    async with db() as session:
+        session.add(LogicalModel(id="default-chat", capabilities=["chat"]))
+        session.add(
+            ProviderEndpoint(
+                logical_model_id="default-chat",
+                provider="openai_compatible",
+                name="mock-endpoint",
+                base_url="http://mock.internal",
+                model_id="mock-chat",
+                api_key_env="HJ_MOCK_KEY",
+                priority=1,
+            )
+        )
+        await session.commit()
     transport = ASGITransport(app=app, raise_app_exceptions=False)
     async with AsyncClient(transport=transport, base_url="https://test") as c:
         yield c
@@ -56,3 +78,25 @@ async def login(client, email: str) -> dict[str, str]:
     resp = await client.post("/v1/auth/sessions", json={"email": email})
     assert resp.status_code == 201, resp.text
     return {"Hunjing-CSRF-Token": resp.json()["csrf_token"]}
+
+
+def _mock_openai_stream():
+    """OpenAI-compatible 流式 mock：真实 SSE 字节流（data: {...} 分块 + usage）。"""
+    import httpx
+
+    sse = (
+        b'data: {"choices":[{"delta":{"content":"\\u4f60\\u597d"}}]}\n\n'
+        b'data: {"choices":[{"delta":{"content":","}}]}\n\n'
+        b'data: {"choices":[{"delta":{"content":"\\u9b42\\u9559"}}]}\n\n'
+        b'data: {"choices":[{"delta":{},"finish_reason":"stop"}]}\n\n'
+        b'data: {"choices":[],"usage":{"prompt_tokens":12,'
+        b'"completion_tokens":6,"total_tokens":18}}\n\n'
+        b'data: [DONE]\n\n'
+    )
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        return httpx.Response(
+            200, headers={"content-type": "text/event-stream"}, content=sse
+        )
+
+    return httpx.MockTransport(handler)