refactor(data): 移除 api_daily 模块并更新文档

- 删除 src/data/api_wrappers/api_daily.py (240行)
- 更新 6 个文档文件，将 daily 表引用替换为 pro_bar
- 同步 README.md 中的因子框架和训练模块示例

BREAKING CHANGE: api_daily 模块已移除，请使用 api_pro_bar 替代

README.md

@@ -36,9 +36,21 @@ ProStock/
 │   │
 │   ├── data/                # 数据获取与存储
 │   │   ├── api_wrappers/    # Tushare API 封装
-│   │   │   ├── api_daily.py         # 日线数据接口
-│   │   │   ├── api_stock_basic.py   # 股票基础信息
-│   │   │   └── api_trade_cal.py     # 交易日历
+│   │   │   ├── api_pro_bar.py            # Pro Bar行情数据接口(主用)
+│   │   │   ├── api_stock_basic.py        # 股票基础信息接口
+│   │   │   ├── api_trade_cal.py          # 交易日历接口
+│   │   │   ├── api_bak_basic.py          # 历史股票列表接口
+│   │   │   ├── api_namechange.py         # 股票名称变更接口
+│   │   │   ├── api_stock_st.py           # ST股票信息接口
+│   │   │   ├── api_daily_basic.py        # 每日指标接口
+│   │   │   ├── api_stk_limit.py          # 涨跌停价格接口
+│   │   │   ├── financial_data/           # 财务数据接口
+│   │   │   │   ├── api_income.py         # 利润表接口
+│   │   │   │   ├── api_balance.py        # 资产负债表接口
+│   │   │   │   ├── api_cashflow.py       # 现金流量表接口
+│   │   │   │   ├── api_fina_indicator.py # 财务指标接口
+│   │   │   │   └── api_financial_sync.py # 财务数据同步调度中心
+│   │   │   └── __init__.py
 │   │   ├── client.py        # Tushare 客户端（含限流）
 │   │   ├── config.py        # 数据模块配置
 │   │   ├── db_manager.py    # DuckDB 表管理和同步
@@ -140,83 +152,123 @@ uv run python -c "from src.data.db_inspector import get_db_info; get_db_info()"
 ### 因子计算
 
 ```python
-from src.factors import FactorEngine, DataLoader, DataSpec
-from src.factors.base import CrossSectionalFactor, TimeSeriesFactor
+from src.factors import FactorEngine
+from src.factors.api import close, ts_mean, cs_rank
 import polars as pl
 
-# 自定义截面因子：PE排名
-class PERankFactor(CrossSectionalFactor):
-    name = "pe_rank"
-    data_specs = [DataSpec("daily", ["ts_code", "trade_date", "pe"], lookback_days=1)]
-    
-    def compute(self, data) -> pl.Series:
-        cs = data.get_cross_section()
-        return cs["pe"].rank()
+# 初始化引擎
+engine = FactorEngine()
 
-# 自定义时序因子：20日移动平均
-class MA20Factor(TimeSeriesFactor):
-    name = "ma20"
-    data_specs = [DataSpec("daily", ["ts_code", "trade_date", "close"], lookback_days=20)]
-    
-    def compute(self, data) -> pl.Series:
-        return data.get_column("close").rolling_mean(window_size=20)
+# 方式1：使用 DSL 表达式注册
+engine.register("ma20", ts_mean(close, 20))
+engine.register("price_rank", cs_rank(close))
 
-# 执行计算
-loader = DataLoader(data_dir="data")
-engine = FactorEngine(loader)
+# 方式2：使用字符串表达式（推荐）
+engine.add_factor("ma20", "ts_mean(close, 20)")
+engine.add_factor("alpha", "cs_rank(ts_mean(close, 5) - ts_mean(close, 20))")
 
-# 计算截面因子
-pe_rank = PERankFactor()
-result1 = engine.compute(pe_rank, start_date="20240101", end_date="20240131")
+# 方式3：从 metadata 查询（需先在 metadata 中定义）
+engine.add_factor("mom_5d")
 
-# 计算时序因子
-ma20 = MA20Factor()
-result2 = engine.compute(ma20, stock_codes=["000001.SZ"], 
-                        start_date="20240101", end_date="20240131")
+# 计算因子
+result = engine.compute(
+    factor_names=["ma20", "price_rank"],
+    start_date="20240101",
+    end_date="20240131"
+)
 
-# 因子组合
-combined = 0.5 * pe_rank + 0.3 * ma20
+# 查看执行计划
+plan = engine.preview_plan("ma20")
 ```
 
 ### 模型训练
 
 ```python
-from src.models import PluginRegistry, ProcessingPipeline
-from src.models.core import PipelineStage
+from src.training import (
+    Trainer,
+    LightGBMModel,
+    DateSplitter,
+    StockPoolManager,
+    NullFiller,
+    Winsorizer,
+    StandardScaler,
+    STFilter,
+    check_data_quality,
+)
+from src.factors import FactorEngine
 import polars as pl
 
-# 创建处理流水线
-pipeline = ProcessingPipeline([
-    PluginRegistry.get_processor("dropna")(),
-    PluginRegistry.get_processor("winsorizer")(lower=0.01, upper=0.99),
-    PluginRegistry.get_processor("standard_scaler")(),
-])
+# 1. 创建模型
+model = LightGBMModel(params={
+    "objective": "regression",
+    "metric": "mae",
+    "num_leaves": 20,
+    "learning_rate": 0.01,
+    "n_estimators": 1000,
+})
 
-# 准备数据
-data = pl.read_csv("features.csv")  # 包含特征和标签
+# 2. 准备因子数据
+engine = FactorEngine()
+engine.add_factor("ma5", "ts_mean(close, 5)")
+engine.add_factor("ma20", "ts_mean(close, 20)")
 
-# 划分训练/测试集
-from src.models.core import WalkForwardSplit
-splitter = WalkForwardSplit(train_window=252, test_window=21)
+# 计算全市场因子
+data = engine.compute(
+    factor_names=["ma5", "ma20", "future_return_5"],
+    start_date="20200101",
+    end_date="20231231"
+)
 
-# 获取 LightGBM 模型
-ModelClass = PluginRegistry.get_model("lightgbm")
-model = ModelClass(task_type="regression", params={"n_estimators": 100})
+# 3. 创建数据处理器
+processors = [
+    NullFiller(feature_cols=["ma5", "ma20"], strategy="mean"),
+    Winsorizer(feature_cols=["ma5", "ma20"], lower=0.01, upper=0.99),
+    StandardScaler(feature_cols=["ma5", "ma20"]),
+]
 
-# 训练循环
-for train_idx, test_idx in splitter.split(data):
-    train_data = data[train_idx]
-    test_data = data[test_idx]
-    
-    # 数据处理
-    X_train = pipeline.fit_transform(train_data.drop("target"))
-    X_test = pipeline.transform(test_data.drop("target"))
-    y_train = train_data["target"]
-    y_test = test_data["target"]
-    
-    # 训练模型
-    model.fit(X_train, y_train)
-    predictions = model.predict(X_test)
+# 4. 创建股票池筛选函数
+def stock_pool_filter(df: pl.DataFrame) -> pl.Series:
+    """筛选小市值股票"""
+    code_filter = (
+        ~df["ts_code"].str.starts_with("300") &  # 排除创业板
+        ~df["ts_code"].str.starts_with("688")    # 排除科创板
+    )
+    return code_filter
+
+pool_manager = StockPoolManager(
+    filter_func=stock_pool_filter,
+    required_columns=["total_mv"],
+)
+
+# 5. 创建过滤器
+st_filter = STFilter(data_router=engine.router)
+
+# 6. 创建数据划分器
+splitter = DateSplitter(
+    train_start="20200101",
+    train_end="20221231",
+    val_start="20230101",
+    val_end="20230630",
+    test_start="20230701",
+    test_end="20231231",
+)
+
+# 7. 创建训练器
+trainer = Trainer(
+    model=model,
+    pool_manager=pool_manager,
+    processors=processors,
+    filters=[st_filter],
+    splitter=splitter,
+    target_col="future_return_5",
+    feature_cols=["ma5", "ma20"],
+)
+
+# 8. 执行训练
+results = trainer.train(data)
+
+# 9. 获取预测结果
+predictions = trainer.get_results()
 ```
 
 ## 核心设计