feat: 本地skill积累更新 - 引擎增强+婚姻时机验证v5/v6+K.N.Rao八参数法

- SKILL.md: 更新至v3.7.4，补充能力描述
- scripts/jyotish_engine.py: 大幅增强计算引擎(+725行)
- scripts/event_prediction_model.py: 事件预测模型重构
- scripts/jaimini.py: Jaimini系统增强
- scripts/dasha_calculator.py: Dasha计算优化
- scripts/varga.py: 分盘计算更新
- jyotish-app/: 前端界面+过境分析功能增强
- references/: 新增3个婚姻时机研究文件
  - verified-patterns-marriage-timing-v5.md
  - verified-patterns-marriage-timing-v6.md
  - rao-marriage-timing-8-parameters.md

scripts/varga.py

@@ -71,6 +71,18 @@ def varga_map(si, pi, div):
 
 def calc_varga(lon, div):
     """计算行星在指定分盘的位置（星座+精确度数+尊贵状态）"""
+    # 2026-05-03修正：D9 Navamsa使用Harmonic(×9)法而非BPHS映射法
+    # 原因：Harmonic法与主流Jyotish软件(Jagannatha Hora等)结果一致
+    # 且在Einstein等名人案例中与人生事件吻合度更高
+    # BPHS映射法(el={0:0,1:9,2:6,3:3})仍保留在varga_map中作为备选
+    if div == 9:
+        nav_lon = (lon * 9) % 360
+        vsi = int(nav_lon / 30) % 12
+        dp = nav_lon - vsi * 30
+        pi = int(dp / (30.0/9))  # approximate pada from harmonic degree
+        r = {'sign': _sn(vsi), 'sign_idx': vsi, 'degree_in_sign': round(dp, 4),
+             'part_index': pi, 'lord': SIGN_LORDS.get(_sn(vsi), ''), 'pada': pi + 1}
+        return r
     si=_si(lon); d=lon-si*30; ps=30.0/div; pi=int(d/ps); dp=d-pi*ps
     vsi=varga_map(si,pi,div)
     r={'sign':_sn(vsi),'sign_idx':vsi,'degree_in_sign':round(dp,4),