fix(backtest): A股适配层—定寸/做空拦截/真实费用/口径统一(Phase1+2)

审计发现包装层系统性失真(2 CRITICAL+7 HIGH)，vnpy底座可信但A股场景未适配：
- C1 定寸: engine.size=N(满仓手数)，策略volume=1手=N股，开平对称(pos归零)
- C2 做空拦截: SHORT+OPEN拒单，long-only，SHORT+CLOSE平多允许
- H3 A股费用: AShareDailyResult重算(佣金保底5元/印花税卖方/过户费沪市)
- H4 收益口径: simple return从balance算(不再用vnpy log return喂empyrical)
- H5+口径: benchmark ffill对齐不缩样本; sizing_shares_per_lot暴露
- H7 退化检测: 零成交/空数据标degenerate不静默done
- H8 task_id: optimize/factor用uuid4(原id()内存地址)
- 静默except改warning

验证: 容器内真实vnpy DoubleMa 600000 2022-2024, total_return 1e-6→42.3%,
end_balance 100万→142万, SHORT+OPEN成交0笔, N=7800股/手.
22 backtest测试全绿(含集成测试), API健康200.

tests/backtest/test_integration_ashare.py

@@ -0,0 +1,83 @@
+"""A股适配层真实集成测试（容器内跑，Mac 本机无法运行）。
+
+需要 vnpy_ctastrategy + quant_trading.db + A 股 K 线数据。
+用法：容器内 `pytest tests/backtest/test_integration_ashare.py -m integration`
+
+验证 Phase 1+2 四项核心断言：
+- C1 定寸：成交金额 ≈ 满仓量级（size=N 方案，volume=1 手=N 股）
+- C2 做空拦截：无 SHORT+OPEN 成交
+- H3 真实费用：end_balance != capital（非空转，费用+盈亏反映在余额）
+- 非噪声：|total_return| > 1e-3
+"""
+import pytest
+
+# 容器内才有 vnpy_ctastrategy；Mac 本机自动 skip（不中断 pytest 全量跑）
+pytest.importorskip("vnpy_ctastrategy")
+
+pytestmark = [pytest.mark.integration]
+
+
+def test_double_ma_600000_2022_2024():
+    """DoubleMa 600000 2022-2024 真实回测验证。
+
+    标记 integration → 仅容器内跑（需 vnpy + quant_trading.db + A 股日线数据）。
+    用 3 年窗口确保 ArrayManager(100) 充分暖机 + 产生足够多 MA 交叉信号
+    （2024H1 窗口太短，仅 111 根 bar 暖机后信号窗口不足，会误判退化）。
+    """
+    from vnpy_ctastrategy.strategies.double_ma_strategy import DoubleMaStrategy
+    from sanguo_backtest.cta_engine import run_cta_backtest
+
+    capital = 1_000_000
+    position_pct = 0.95
+
+    result = run_cta_backtest(
+        strategy_class=DoubleMaStrategy,
+        symbol="600000",
+        params={"fast_window": 5, "slow_window": 10},
+        start="2022-01-01",
+        end="2024-12-31",
+        cfg=None,  # cta_engine 内部 load_config
+        db_path="/tmp/test_integration_ashare.db",
+        benchmark="hs300",
+        capital=capital,
+        position_pct=position_pct,
+    )
+
+    # 基本成功检查
+    assert result.status in ("done", "degenerate"), f"回测失败: {result.error_msg}"
+    assert result.status == "done", f"回测退化（不应退化）: {result.statistics.get('degenerate_reason')}"
+
+    stats = result.statistics
+    trades = result.trades
+
+    # H3: end_balance != capital（非空转——有费用+盈亏）
+    end_balance = stats.get("end_balance")
+    assert end_balance is not None, "statistics 缺 end_balance"
+    assert abs(end_balance - 1_000_000) > 1.0, f"end_balance={end_balance} 与 capital 几乎相同（空转）"
+
+    # C2: 做空拦截——无 SHORT+OPEN
+    if trades is not None and not trades.empty:
+        short_opens = trades[
+            (trades["direction"].str.contains("SHORT"))
+            & (trades["offset"].str.contains("OPEN"))
+        ]
+        assert len(short_opens) == 0, f"存在 SHORT+OPEN 成交（做空未拦截）: {short_opens}"
+
+    # C1: 定寸生效——成交金额 ≈ 满仓量级（size=N 方案：volume=1 手，turnover=1*N*price）
+    sizing_shares_per_lot = stats.get("sizing_shares_per_lot", 0)
+    if trades is not None and not trades.empty and sizing_shares_per_lot > 0:
+        first_trade = trades.iloc[0]
+        turnover = first_trade["volume"] * sizing_shares_per_lot * first_trade["price"]
+        assert turnover > capital * position_pct * 0.5, (
+            f"首笔成交金额={turnover:.0f} 未达满仓量级 "
+            f"(capital={capital} pct={position_pct} N={sizing_shares_per_lot})"
+        )
+
+    # 非噪声：|total_return| > 1e-3
+    total_return = stats.get("total_return")
+    if total_return is not None:
+        assert abs(total_return) > 1e-3, f"|total_return|={abs(total_return)} <= 1e-3（噪声）"
+
+    # H3: 费用可见——statistics 含 stamp_duty 或 commission > 0
+    total_commission = stats.get("total_commission", 0)
+    assert total_commission > 0, f"total_commission={total_commission}（费用未计入）"