feedback7 R22-FIELDS: 耀骑士临光's 卡西米尔 redeploy reduction (garrison 144) has no 5 % floor — the multiplier clamps at 0 like every other redeploy multiplier, so a knock-out timer can reach 0 s (GitHub #370); golden: unchanged

GitHub #370 (FrogThai): 「耀骑士临光特性赋予身前干员减再部署有上限，最多减至5秒」. Cause: content/garrisons/battle.js
applyReader, act1autochess_gar_eff_respawnTimeByBond, wrote redeployMul = max(0.05, 1 + respawn_time × steps), so the
timer stopped at 5 % of the base time — from 192 卡西米尔 layers for 144_a (raw 0.04) and from 96 for 144_b. Measured on
a 100 s base (notes repro-370.mjs): 144_a 192 / 300 layers → 5.00 s / 5.00 s, 144_b 96 / 192 / 300 → 5.00 s each.
Official: data/garrisons.json garrison_144_a bb { divide_num 3, respawn_time −0.015 } 「每叠加3层，本干员再部署时间-1.5%」,
144_b −0.03 「-3%」 — no minimum key; PRTS 耀骑士临光 has no 卫戍 note on a minimum. Now max(0, …) as items/battle.js,
bands, the 协防 bond and svrash; deploy.js already clamps the timer at 0: 144_a 192 / 300 layers → 4.00 s / 0 s,
144_b 96 / 192 / 300 → 4.00 s / 0 s / 0 s. A 0 s timer brings the operator back on the next tick, no error (notes
probe-370-redeploy.mjs). The ASPD part (garrison 159 / 160) is untouched. Patch from the Grok triage
(scratch-i1b/370.patch), the test extended with 96 layers and the 0 s return.

Tests: test/content/garrisons_battle.test.js 26/26 — the new #370 test (144_a / 144_b at 9, 96, 192, 300 layers: the
multiplier, the knock-out timer, the 0 s return) fails on the parent.

server/sim/content/garrisons/battle.js

@@ -400,7 +400,10 @@ function applyReader(battle, it) {
       else battle.removeBuff(u, key);
       break;
     case 'act1autochess_gar_eff_respawnTimeByBond':
-      if (k > 0) S.passiveBuff(battle, u, key, { redeployMul: Math.max(0.05, 1 + S.num(it.bb.respawn_time, 0) * k) });
+      // 耀骑士临光's 144: −1.5 % / −3 % per 3 卡西米尔 layers with no minimum in the blackboard (divide_num, respawn_time
+      // only) — clamped at 0 like every other redeploy multiplier, so a knock-out timer can reach 0 s (GitHub #370;
+      // until 0.2.1 a 0.05 floor kept it at 5 % of the base time from 192 layers, 96 for the elite)
+      if (k > 0) S.passiveBuff(battle, u, key, { redeployMul: Math.max(0, 1 + S.num(it.bb.respawn_time, 0) * k) });
       else battle.removeBuff(u, key);
       break;
     case 'act2autochess_gar_eff_attrByBond_add_onstart': {

test/content/garrisons_battle.test.js

@@ -465,6 +465,33 @@ test('ADD_BOND 耀骑士临光 145 / 160 (real chess): self + front get 144 (red
   }
 });
 
+test('respawnTimeByBond 144 has no 5% floor (#370): redeployMul is max(0, 1 + respawn_time * steps)', () => {
+  for (const [gid, rt] of [['garrison_144_a', -0.015], ['garrison_144_b', -0.03]]) {
+    assert.equal(GR(gid).bb.respawn_time, rt);
+    assert.deepEqual(Object.keys(GR(gid).bb).sort(), ['divide_num', 'respawn_time']);
+    // 96 / 192 layers: the old 0.05 floor bound the elite / the normal form (raw 0.04); 300: below zero
+    for (const layers of [9, 96, 192, 300]) {
+      const k = Math.floor(layers / 3);
+      const want = Math.max(0, 1 + rt * k);
+      const h = battle({
+        units: [{ id: 'op', g: [gid], row: 10, col: 4, o: { stats: { respawnTime: 100 } } }],
+        bonds: { kazimierzShip: B(layers) },
+      });
+      const u = h.unit('op');
+      approx(u.findBuff(`gar:${gid}`).mods.redeployMul, want, `${gid} @ ${layers}`);
+      h.b.dealDamage(null, u, { amount: 1e9, type: 'true' });
+      assert.equal(u.alive, false);
+      approx(u.respawnAt - u.deathAt, 100 * want, `${gid} timer @ ${layers}`);
+      if (want === 0) {
+        h.step(2);
+        assert.ok(u.alive && u.deployed, `${gid} @ ${layers}: a 0 s timer brings it straight back`);
+        assert.equal(h.b.errorCount, 0);
+      }
+    }
+    cover(gid);
+  }
+});
+
 // ---------------------------------------------------------------------------------------------------------------------
 // 魔王