feat(dashboard): show AI pick evidence gaps

tests/test_frontend_v2_assets.py

@@ -200,6 +200,8 @@ def test_ai_product_pick_agent_uses_real_competitor_data_and_dashboard_action():
     assert "evidence_quality" in agent_source
     assert "opportunity_score" in agent_source
     assert "margin_rate" in agent_source
+    assert "missing_evidence" in agent_source
+    assert "confidence_band" in agent_source
     assert "_supersede_old_picks" in agent_source
     assert "status = 'superseded'" in agent_source
     assert "{date_col}::date" in agent_source
@@ -215,6 +217,15 @@ def test_ai_product_pick_agent_uses_real_competitor_data_and_dashboard_action():
     assert "完成後會重算 AI 挑品清單" in route_source
     assert "match_rate" in route_source
     assert "product_pick_count" in route_source
+    dashboard_route_source = (ROOT / "routes/dashboard_routes.py").read_text(encoding="utf-8")
+    dashboard_template = (ROOT / "templates/dashboard_v2.html").read_text(encoding="utf-8")
+    assert "model_footprint" in dashboard_route_source
+    assert "_ai_pick_evidence_fields" in dashboard_route_source
+    assert "avg_evidence_quality" in dashboard_route_source
+    assert "top_missing_evidence" in dashboard_route_source
+    assert "證據 {{ item.ai_pick.evidence_quality" in dashboard_template
+    assert "dashboard-ai-evidence-chip" in dashboard_template
+    assert "EVIDENCE GAP" in dashboard_template
 
     scheduler_source = (ROOT / "scheduler.py").read_text(encoding="utf-8")
     run_scheduler_source = (ROOT / "run_scheduler.py").read_text(encoding="utf-8")