feedback4: a sleeping enemy cannot be blocked and takes no block slot — 沉睡 = 无法行动+无敌+不可阻挡 (GitHub #140; the owner's decision of 2026-10-06)

- GitHub #140: a player tested 瑕光's S2 in the official mode — she slept 4 enemies and still blocked the next ones,
  after the skill no more were (「被沉睡的敌人是不占用阻挡数的」). PRTS 异常效果: SLEEPING 沉睡 = 「无法行动+无敌+不可阻挡」,
  BLOCK_FREE 不可阻挡 = 「无法阻挡/被阻挡，自动解除阻挡」. Until now an enemy that fell asleep while blocked stayed blocked
  and kept its blocker's slot.
- Battle.applyStatus releases an enemy that falls asleep (as for 浮空 / 恐惧 / 不可阻挡), Battle._checkBlock never blocks a
  sleeper, and ai.js updateEnemy lets a plain-buff sleeper go on its next update. The sleeper stays put; awake, it is
  blocked again only by a blocker with a free slot, else it walks on. A stun, a freeze or a refused sleep keeps the
  block; sleeping operators are unchanged.
- Readers of the block follow: Battle.blockedTargets (blocked-first targeting, the DEFAULT trigger, hitAllBlocked)
  drops the sleeper — of the hitSleep attackers only 瑕光 blocks, and her 仁慈 reads her range; a 要塞 holding only a
  sleeper no longer stalls; “庞贝”'s self-blast clock restarts when a sleep ends its block [ASSUMED]; the snapshot's
  blocked bit follows.
- Tests: new test/sim/feedback4-sleep-unblock.test.js (5 cases, each fails without the change: the slot frees and the
  next enemy is blocked, the sleeper stays put; awake with no room it walks on; never blocked while asleep, re-blocked
  on waking beside a free blocker; a plain-buff sleep; the real 瑕光 S2). 1274 targeted tests (blocking, statuses,
  小满 / 瑕光 / 缇缇 kits, garrisons, professions, enemies, kits t1–t6), docs-consistency and GOLDEN_FULL pass.
- Docs: DESIGN §24.9 (and the §24 intro, §5.5 Blocking, §22.8's release list), SIM.md (Blocking, the `sleep` row,
  plain buffs), the buffs.js / Battle.js / ai.js / 瑕光 kit comments.
- Golden (every moved scenario sleeps blocked enemies; DESIGN §24.9 gives each move): roster-003 / roster-005 (小满 S2:
  82 kills instead of 83 / cleared at 136.73 s instead of the 140 s timeout), roster-013 (缇缇 S3: only the blocked
  count and the snapshot hashes), roster-032 (缇缇 S1: 7 knock-outs instead of 5), roster-038 (elite 缇缇 S2: ends at
  197.43 s), roster-039 (瑕光 S2 normal + elite: 41 of 52 killed instead of 47 — the sleepers walk on when they wake),
  roster-042 (缇缇 S2: ends at 161.17 s instead of 166.97 s), bond-maniShip-high (小满: only the blocked count and the
  snapshot hashes), coop4-FUNNY-6 (ai_0's 缇缇: layers and boss damage from round 9), coop4-ABYSS-7 (ai_2's 小满:
  rounds 11–12), coop2-NORMAL-8-serverrun (ai_1's 缇缇: one more leak in round 10, the team eliminated in round 13
  instead of winning). fields unchanged.

server/sim/buffs.js

@@ -58,7 +58,9 @@ export const STATUS = Object.freeze({
   // 寒冷: ASPD −30. A second cold while this one remains applies 冻结 for max(remaining, incoming)
   // (Battle.applyStatus; PRTS 术语释义 寒冷 「持续时间取双方之中最高」).
   cold: { flags: { cold: true }, mods: { aspd: COLD_ASPD } },
-  // 沉睡: 无敌且无法行动 (untargetable, takes no damage — except from `hitSleep` attackers — and blocks nothing)
+  // 沉睡: 无敌且无法行动 (untargetable, takes no damage — except from `hitSleep` attackers — and blocks nothing); PRTS 异常效果
+  // SLEEPING = 无法行动+无敌+不可阻挡: an enemy asleep is not blocked either — its blocker lets go and the slot frees, it
+  // stays put until it wakes (the `sleep` flag: Battle._checkBlock / applyStatus, ai.js updateEnemy; DESIGN §24.9)
   sleep: { flags: { sleep: true, noBlock: true }, immune: 'sleep' },
   // value = fraction of speed removed (0.8 ⇒ moveMul 0.2). Default 0.5.
   slow: { mods: (v) => ({ moveMul: 1 - clamp01(v ?? 0.5) }), valued: 0.5 },