feat(harness): add Action Recognition training experiment

- PerCategoryPoolStrategy: filter test pool by task_types
- RunConfig: add run_holdout_eval toggle (default true)
- load_config: fix YAML task_types list-to-tuple conversion
- Runner: conditionally skip _holdout_four_way when disabled
- CLI: add --no-run-holdout-eval flag
- New config/train_action_recognition.yaml (3 epochs, per_category)
- New scripts/train_action_recognition.sh (baseline + seed + train)

research-wiki/graph/edges.json

@@ -145,6 +145,21 @@
       "id": "plan:2026-07-11-question-gen-v2",
       "label": "出题管线 v2 实现计划",
       "type": "plan"
+    },
+    {
+      "id": "design:per-category-pool-strategy",
+      "label": "Per-Category Pool Strategy 设计",
+      "type": "design"
+    },
+    {
+      "id": "plan:per-category-pool-strategy",
+      "label": "Per-Category Pool Strategy 实现计划",
+      "type": "plan"
+    },
+    {
+      "id": "plan:action-recognition-training",
+      "label": "Action Recognition 单题型首次训练实验计划",
+      "type": "plan"
     }
   ],
   "links": [
@@ -259,6 +274,20 @@
       "relation": "implements",
       "evidence": "Spec-3 设计的实现计划",
       "added": "2026-07-12T02:57:39.569968+00:00"
+    },
+    {
+      "source": "plan:per-category-pool-strategy",
+      "target": "design:per-category-pool-strategy",
+      "relation": "implements",
+      "evidence": "实现 PoolStrategy Protocol + PerCategoryPoolStrategy 设计",
+      "added": "2026-07-13T02:26:58.551190+00:00"
+    },
+    {
+      "source": "plan:action-recognition-training",
+      "target": "design:action-recognition-training",
+      "relation": "implements",
+      "evidence": "计划实现设计文档中的 2 处代码修改 + 实验配置 + 训练脚本",
+      "added": "2026-07-14T04:50:15.986586+00:00"
     }
   ]
 }