feat(pchome): close freshness and multi-market decision gaps

tests/test_elephant_alpha_engine.py

@@ -146,7 +146,7 @@ def test_competitor_db_evidence_actions_are_concrete():
     )
 
     assert actions == [
-        "[SKU-1] 測試商品｜MOMO $1,200 vs PChome $990 (+17.5%)｜每件價差 NT$ 210｜證據：高信心同款 / 總價可比 / 可直接價格告警 / score 0.84｜建議人工確認 PChome identity_v2 後評估跟價或促銷｜PChome D123456"
+        "[SKU-1] 測試商品｜MOMO $1,200 vs PChome $990 (+17.5%)｜每件價差 NT$ 210｜證據：高信心同款 / 總價可比 / 可直接價格告警 / score 0.84｜建議 AI 自動驗證確認 PChome identity_v2 後評估跟價或促銷｜PChome D123456"
     ]
 
 
@@ -279,9 +279,10 @@ def test_price_review_decision_envelope_blocks_auto_execution():
 
     assert envelope["decision_type"] == "price_decision_review"
     assert envelope["source_agent"] == "elephant_alpha"
-    assert envelope["recommended_action"]["requires_hitl"] is True
+    assert envelope["recommended_action"]["requires_hitl"] is False
+    assert envelope["recommended_action"]["requires_ai_exception"] is True
     assert envelope["guardrails"]["can_auto_execute"] is False
-    assert envelope["guardrails"]["blocked_reason"] == "price decisions require HITL; execution_plan skipped"
+    assert envelope["guardrails"]["blocked_reason"] == "price decisions require AI 例外決策; execution_plan skipped"
     assert envelope["guardrails"]["data_quality"] == "complete"
 
 
@@ -701,7 +702,8 @@ def test_human_escalation_decision_envelope_blocks_auto_execution():
     assert envelope["decision_type"] == "ea_escalation"
     assert envelope["guardrails"]["can_auto_execute"] is False
     assert envelope["guardrails"]["data_quality"] == "partial"
-    assert envelope["recommended_action"]["requires_hitl"] is True
+    assert envelope["recommended_action"]["requires_hitl"] is False
+    assert envelope["recommended_action"]["requires_ai_exception"] is True
     assert envelope["trace"]["provider"] == "elephant_alpha"