test: auto-generate 60-rule yoga test suite from 60-chart benchmark

- yoga_test_suite.json: 60 rules x True/False test cases generated
  from the 60-chart benchmark (40 rules have both True+False)
- run_yoga_test_suite.py: test runner, 97/100 passed, 3 known FN
  residuals (bahu_puthra, dehasthoulya, grihanasa)
- Inspired by PyJHora's hardcoded _from_planet_positions methodology

tests/run_yoga_test_suite.py

@@ -0,0 +1,44 @@
+"""
+自动化 Yoga 测试运行器 — 从 yoga_test_suite.json 加载用例并批量验证
+"""
+import json, sys, os
+SKILL_DIR = '/Users/wuyongnaren/.workbuddy/skills/jyotish-vedic-astrology'
+sys.path.insert(0, os.path.join(SKILL_DIR, 'scripts'))
+RULES_PATH = os.path.join(SKILL_DIR, 'references', 'yoga_rules.json')
+from yoga_engine import YogaEngine
+engine = YogaEngine(RULES_PATH)
+
+def detect(rule_id, planets, asc, ctx=None):
+    return any(r.get('rule_id') == rule_id for r in engine.detect(planets, asc, context=ctx))
+
+suite_path = os.path.join(SKILL_DIR, 'tests', 'yoga_test_suite.json')
+with open(suite_path) as f:
+    suite = json.load(f)
+
+passed = 0
+failed = 0
+failures = []
+
+print(f'Yoga Test Suite: {len(suite)} rules')
+print('=' * 60)
+
+for entry in suite:
+    rid = entry['rule_id']
+    rid_short = rid[-30:] if len(rid) > 30 else rid
+    
+    for mode in ['true', 'false']:
+        if mode not in entry: continue
+        d = entry[mode]
+        expected = (mode == 'true')
+        result = detect(rid, d['planets'], d['asc'], d.get('context'))
+        
+        if result == expected:
+            passed += 1
+        else:
+            failed += 1
+            chart = d.get('chart', '?')
+            failures.append(f'  FAIL: {rid_short:30s} {mode:5s} chart={chart} expect={expected} got={result}')
+
+for f in failures:
+    print(f)
+print(f'\nResults: {passed} passed, {failed} failed ({passed+failed} total)')