docs: 排序修正+轮次53——把压在最新轮次之上的"轮次39"块移回它的位置；§9.2 记分板换到 K16+pre 现状

排序：倒序约定此前只在各轮之间成立，旧的"轮次 39"及其 4 个续篇块卡在候选表下面、
轮次 52 上面 ⇒ 打开日志第一眼是 39，看起来像"没再更新"。整块 67 行移到 轮次38 之前，
纯移动不改内容（字节数 188223 不变，LF 保持）。

轮次53（seed=1 逐题配对）：pre 在 §9.2 自己的 1.3x 口径上复现——s0 22新增/5失去 p=0.0015，
s1 17/6 净+11 p=0.035，合并 39/11 p=9.0e-5 ⇒ "打分地平线=执行地平线"确实买到时限这一栏。
门控20mm 在 s1 只有 12/11 净+1 (p=1.0)，合并 +17 p=0.024 ⇒ 修正轮次50 说过头的"塌了"，
准确说法是"门控吃掉约四成时限收益，损失集中在 seed=1"。

记分板更新（最好合法口径 = inc_cap+K16+pre，轮次51）：碰撞 0 / 超VMAX 0 达标；
可达到达 0.5967(差 38.3pp，相对控制净+11 题 p=0.061 未确立)；时限内 0.3594(唯一被配对证明)；
行程比 2.414(超 1.3 倍)；brake_rate mean 7.28% 但 p50=0.0%、p90 27.4%(长尾, stuck_share 17.2%)。

docs/experiment_round53_ctl_ex_s1.json

@@ -0,0 +1,110 @@
+{
+  "config": {
+    "ckpt": "ml/runs/base_dag2_dag1/best.pt",
+    "samples": "ml/data/samples",
+    "tasks": null,
+    "split": "val",
+    "n": 256,
+    "K": 8,
+    "commit": 2,
+    "fb": 2.2,
+    "ki": 0.0,
+    "fb_scale": 0.0,
+    "fb_near": 0.2,
+    "fb_fallback": 0.0,
+    "kdecay": 0.9,
+    "kimax": 0.5,
+    "fb_every": 1,
+    "safety_try": 0,
+    "sel_w": 0.1,
+    "dls_safety": 1,
+    "fb_clear": 0.0,
+    "fb_cmin": 0.3,
+    "sub_refresh": 0,
+    "fb_block_gain": 0.5,
+    "dump": "ml/runs/round53_ctl_ex_s1_rows.json",
+    "out": "docs/experiment_round53_ctl_ex_s1.json",
+    "trace": null,
+    "select": "endpoint",
+    "replan": "empty",
+    "safety": 1,
+    "partial": 1,
+    "inc_cap": 1,
+    "gap_margin": 0.0,
+    "cert_batch": 1,
+    "steps": 1,
+    "z_scale": 1.0,
+    "len_w": 0.0,
+    "tie": 0,
+    "pre_h": 0,
+    "pre_near": 0.0,
+    "backoff": 0,
+    "max_sec": null,
+    "seed": 1,
+    "dls": "analytic",
+    "device": "cuda"
+  },
+  "batch_size": 256,
+  "batch_seed_rule": "same seed per batch",
+  "checkpoint_sha256": "f7f686fd8ee92fb6948fbcc4ee954636065ad87bb54f661c4f8f575bc16c2ce1",
+  "tasks_sha256": "249bd93457825cda98c59e64d9a096d3473c69aeb3553de1dea6d9e56be71772",
+  "kin_fp": "2ebcbf17",
+  "act_scale": 1.0,
+  "metrics": {
+    "n": 256,
+    "n_ref": 256,
+    "success": 0.53125,
+    "collision_free_success": 0.53125,
+    "time_ok_1p3": 0.296875,
+    "collision_rate": 0.0,
+    "collision_steps_mean": 0.0,
+    "travel_ratio_median": 2.636941432952881,
+    "time_mean_s": 2.4600000381469727,
+    "brakes_mean": 19.265625,
+    "start_fail": 0.0,
+    "goal_ratio_2s": 0.2750342786312103,
+    "err_fail_med": 214.6193084716797,
+    "err_success_med": 0.22276456654071808,
+    "dls_share": 0.62890625,
+    "time_budget": "teacher×2(≥3 s)",
+    "speed_ratio_p50": 1.0000016689300537,
+    "speed_ratio_p95": 1.0000101327896118,
+    "speed_ratio_max": 1.0000101327896118,
+    "over_vmax_rate": 0.0,
+    "accel_ratio_p95": 1.0000016689300537,
+    "accel_ratio_max": 1.0000101327896118,
+    "reverse_p50": 0.25339367985725403,
+    "reverse_mean": 0.2575649917125702,
+    "reverse_moving_p50": 0.3117443919181824,
+    "moving_share_p50": 0.9636363387107849,
+    "brake_rate_mean": 0.09879808127880096,
+    "brake_rate_p50": 0.0051282052882015705,
+    "brake_rate_p90": 0.3384615480899811,
+    "stuck_share": 0.25390625,
+    "n_plans_mean": 195.0,
+    "success_reachable": 0.5596708059310913,
+    "by_cat": {
+      "A": [
+        128,
+        96
+      ],
+      "B": [
+        74,
+        21
+      ],
+      "C": [
+        32,
+        12
+      ],
+      "D": [
+        13,
+        0
+      ],
+      "Dr": [
+        9,
+        7
+      ]
+    }
+  },
+  "simulation_wall_s": 290.3290144000057
+}