fix(report): FE 问卷键名双侧兼容 + worker 持久事件循环（live E2E 两处修复）

FE↔后端联调实测暴露并修复：
1. 问卷键名三处错位（FE height_cm/weight_kg/psych_style vs 文档
   height/weight/psyche_style）——vision._bmi 多键取值、diagnosis 维度
   装配与规则上下文双键或运算，两侧谁改键名都不再炸
2. celery worker asyncio.run 每投递起新 loop + 模块级引擎缓存 →
   asyncpg 连接跨 loop（同进程第 3 个报告任务必现 RuntimeError:
   attached to a different loop；集成测试单循环掩盖）。新增
   apps/worker/loop.py 持久 loop，report/ping 两任务接入

E2E 终态： psyche_style/height_weight 恢复、BMI 体型确定值、ok 26/32、
真实 LLM 文案通过校验 degraded=False。

apps/worker/loop.py

@@ -0,0 +1,26 @@
+"""Celery worker 的持久事件循环。
+
+``asyncio.run`` 每次投递都新建并随即关闭事件循环，而
+``packages.shared.db`` 的引擎是模块级缓存——asyncpg 连接绑定在创建它的
+loop 上，后续投递从池里拿到旧 loop 的连接就会
+``RuntimeError: attached to a different loop``（live E2E 实测：同进程第
+三个报告任务必现，集成测试单循环跑掩盖了它）。
+
+worker 进程内复用一个 loop，让连接池与 loop 生命周期一致，问题从根上
+消失。prefork 模型下每个子进程各有一份模块状态，互不影响。集成测试
+不受影响：它们在 pytest 自己的 loop 里直调任务内芯，不经过本模块。
+"""
+
+import asyncio
+from collections.abc import Coroutine
+from typing import Any
+
+_loop: asyncio.AbstractEventLoop | None = None
+
+
+def run(coro: Coroutine[Any, Any, Any]) -> Any:
+    """在 worker 进程唯一的持久 loop 上执行协程并返回结果。"""
+    global _loop
+    if _loop is None or _loop.is_closed():
+        _loop = asyncio.new_event_loop()
+    return _loop.run_until_complete(coro)

apps/worker/report_tasks.py

@@ -14,15 +14,15 @@ worker 侧 LLM 网关装配：apps.api.deps 属 HTTP 层，worker 不得 import
 
 from __future__ import annotations
 
-import asyncio
 from datetime import UTC, datetime
 from decimal import Decimal
-from typing import Any
+from typing import Any, cast
 
 from sqlalchemy import func, select
 from sqlalchemy.ext.asyncio import AsyncSession
 from sqlalchemy.orm.attributes import flag_modified
 
+from apps.worker import loop as worker_loop
 from apps.worker.celery_app import celery_app
 from packages.billing.service import BillingService
 from packages.domain.task.state import TERMINAL_STATUSES, TaskStatus
@@ -279,7 +279,7 @@ def generate_report(self: Any, task_id: int) -> dict[str, Any]:
     """
     bind_context(task_id=task_id)
     try:
-        return asyncio.run(_generate_report(task_id))
+        return cast("dict[str, Any]", worker_loop.run(_generate_report(task_id)))
     except StageFailureError as exc:
         logger.warning(
             "report_stage_failed",
@@ -293,5 +293,5 @@ def generate_report(self: Any, task_id: int) -> dict[str, Any]:
             logger.warning("report_generate_retry", task_id=task_id, error=str(exc))
             raise self.retry(exc=exc, countdown=10) from exc
         logger.exception("report_generate_failed_after_retry", task_id=task_id)
-        asyncio.run(_fail_report(task_id, f"{type(exc).__name__}: {exc}"))
+        worker_loop.run(_fail_report(task_id, f"{type(exc).__name__}: {exc}"))
         raise

apps/worker/tasks.py

@@ -9,9 +9,9 @@ render) follow exactly this shape and additionally call
 
 from __future__ import annotations
 
-import asyncio
-from typing import Any
+from typing import Any, cast
 
+from apps.worker import loop as worker_loop
 from apps.worker.celery_app import celery_app
 from packages.domain.task.state import TaskStatus
 from packages.shared.context import bind_context
@@ -61,9 +61,9 @@ async def _fail_ping(task_id: int, error: str) -> None:
 def ping_task(self: Any, task_id: int, echo: str | None = None) -> dict[str, Any]:
     bind_context(task_id=task_id)
     try:
-        return asyncio.run(_run_ping(task_id, echo))
+        return cast("dict[str, Any]", worker_loop.run(_run_ping(task_id, echo)))
     except Exception as exc:
         logger.exception("ping_task_failed", task_id=task_id)
         # Real tasks additionally refund the quota reservation here (FR-BILL-05).
-        asyncio.run(_fail_ping(task_id, f"{type(exc).__name__}: {exc}"))
+        worker_loop.run(_fail_ping(task_id, f"{type(exc).__name__}: {exc}"))
         raise