日K加载提速：adj_factor 覆盖索引 + 冷路径并发 + 热路径进程内缓存直返

- adj_factor 覆盖索引 (ts_code,trade_date) INCLUDE (adj_factor)：根治堆碎片化
  （单股 6516 行散 6516 块，满载时位图堆扫 1.5s+），Index Only Scan ~2ms；
  因子查询全部改 2 列投影，lag() 窗口只取变点（6516→32 行）
- preview 冷路径 2 会话 2 波并发，信息卡合并为 LEFT JOIN LATERAL 一条
- 鉴权会话 60s 进程内缓存（登出/全端登出即时失效），全站请求省 ~80ms
- pvj/chipsj/stocksj/facetsj 存 model_dump_json 原串直返（与 response_model
  字节一致），热路径 230-2190ms → 1-2ms；get_version 本地缓存+bump 即时可见
- 连接池 10+20；smoke_test 适配鉴权缓存

backend/app/models.py

@@ -9,7 +9,7 @@ Candle 表设计与 TimescaleDB hypertable 完全兼容：将来在目标 PG 库
 """
 from datetime import date, datetime
 
-from sqlalchemy import BigInteger, Boolean, Date, DateTime, Float, ForeignKey, Integer, String, Text, UniqueConstraint
+from sqlalchemy import BigInteger, Boolean, Date, DateTime, Float, ForeignKey, Index, Integer, String, Text, UniqueConstraint
 from sqlalchemy.orm import Mapped, mapped_column, relationship
 
 from .db import Base
@@ -136,11 +136,14 @@ class AdjFactor(Base):
 
     id: Mapped[int] = mapped_column(Integer, primary_key=True, autoincrement=True)
     trade_date: Mapped[datetime] = mapped_column(DateTime, index=True)
-    ts_code: Mapped[str] = mapped_column(String(12), index=True)
+    ts_code: Mapped[str] = mapped_column(String(12))
     adj_factor: Mapped[float] = mapped_column(Float)
 
     __table_args__ = (
         UniqueConstraint("ts_code", "trade_date", name="uq_adj_code_date"),
+        # 覆盖索引：因子查询只取 (trade_date, adj_factor) 两列时走 Index Only Scan，
+        # 免堆访问（adj_factor 堆碎片化严重，见 alembic/versions/20260902_01）
+        Index("ix_adj_code_date_cover", "ts_code", "trade_date", postgresql_include=["adj_factor"]),
     )