feat(backtest): 接真实 CTA 策略跑通端到端回测（DoubleMaStrategy on 600000）

修复 cta_engine 在真数据上的多个 bug（Phase 2 未在真数据验证）：
- interval "1d" -> "d"（vnpy Interval.DAILY.value）
- capital 0 -> 1_000_000（0 致首笔交易即爆仓，统计全 0）
- statistics 改用 calculate_statistics(df)（旧代码误用 calculate_result 拿 DataFrame）
- statistics JSON-safe（vnpy 可能含 Timestamp）
- test_cta_engine mock 匹配新流程（calculate_statistics 返回统计字典）

验证：diag_cta.py 真实回测 DoubleMaStrategy on 600000 (2024H1, 111 天)
→ 真实统计 total_return -0.017% / sharpe -1.03 / max_drawdown -2.17 / 1 trade
容器 79 tests passed。

tests/backtest/test_cta_engine.py

@@ -15,9 +15,11 @@ class TestRunCtaBacktest:
         mock_strategy_class = Mock()
         mock_strategy_class.__name__ = "TestStrategy"
 
-        # Mock BacktestingEngine
+        # Mock BacktestingEngine — calculate_result() returns daily_df (DataFrame),
+        # calculate_statistics(df) returns the stats dict (vnpy API, matches cta_engine)
         mock_engine = MagicMock()
-        mock_engine.calculate_result.return_value = {
+        mock_engine.calculate_result.return_value = MagicMock(name="daily_df")
+        mock_engine.calculate_statistics.return_value = {
             "total_return": 0.15,
             "sharpe_ratio": 1.2,
             "max_drawdown": -0.08,
@@ -68,6 +70,7 @@ class TestRunCtaBacktest:
         mock_engine.load_data.assert_called_once()
         mock_engine.run_backtesting.assert_called_once()
         mock_engine.calculate_result.assert_called_once()
+        mock_engine.calculate_statistics.assert_called_once()
 
     def test_run_cta_backtest_handles_exception(self, temp_db_path):
         """Test that exceptions during backtest are handled properly."""