feat: 因果链三层验证机制(P2) — 数据验证+人工确认+状态机

- kpi_causality 加列: source_type/verify_status/verified_at/verified_by/entity_id(回填)
- 核心服务: app/services/causality_verification.py (Pearson+滞后对齐+状态机)
- 数据验证脚本: scripts/correlation-check.py (月度cron, 输出JSON报告)
- API: create/update支持source_type, GET /verify-status, PUT /{id}/verify(人工确认)
- 全部端点按entity_id账套隔离, kpi/{id}/network/simulate补跨企业校验
- 前端: KPIDetail因果链页显示验证状态徽标(数据证实/存疑/待检)
- 测试: test_causality_verification.py 37用例 + 原因果链测试全过(59个)
- 50条因果链首轮验证: 1数据证实(#37渠补率到净利润lag1 r=-0.89), 4存疑, 45待检(数据不足)

backend/app/models/__init__.py

@@ -292,12 +292,17 @@ class KPICausality(Base):
     """KPI因果链 — 记录KPI间的因果关系"""
     __tablename__ = "kpi_causality"
     id = Column(Integer, primary_key=True, index=True)
+    entity_id = Column(Integer, default=1, comment="企业ID (多租户隔离 2026-08-27)")
     source_kpi_id = Column(Integer, ForeignKey("kpi_definitions.id"), nullable=False, comment="源KPI（因）")
     target_kpi_id = Column(Integer, ForeignKey("kpi_definitions.id"), nullable=False, comment="目标KPI（果）")
     strength = Column(Float, default=0.5, comment="影响强度 0~1")
     lag_months = Column(Integer, default=1, comment="滞后期（月）")
     formula = Column(String(500), nullable=True, comment="影响公式描述")
     direction = Column(String(10), default="positive", comment="positive/negative 正向/负向影响")
+    source_type = Column(String(20), default="manual", comment="建链来源 AI_suggested/manual/imported (2026-08-27 验证机制)")
+    verify_status = Column(String(20), default="pending", comment="验证状态 pending/data_verified/human_verified/disputed (2026-08-27)")
+    verified_at = Column(DateTime, nullable=True, comment="验证时间")
+    verified_by = Column(String(50), nullable=True, comment="验证人/AI/脚本")
     created_at = Column(DateTime, server_default=func.now())
     updated_at = Column(DateTime, server_default=func.now(), onupdate=func.now())