feedback1 integrate: the prep dispatcher takes every equipped item before the first one runs (follow-up to the PR #2 fix)

Step 5 read each holder's items when the walk reached that holder, so an item equipped meanwhile onto a holder not yet
walked ran in the same dispatch and an item moved to a later holder ran twice (latent: no shipped item does either).
Every [holder, item] pair is now taken before the first item runs, as the header and META already said.

server/match/effectsMeta.js

@@ -23,10 +23,12 @@
 // ctx.triggerGarrisons(uid, eventType) re-runs another piece's garrisons (铃兰, "触发…的获得时效果", 特质相同).
 // Items: onEquip / onArt / onDestroy go to the item's own handler only; every other hook runs for items equipped on
 // owned chess (ev.source.holder). That step walks a snapshot — the owned chess and each holder's items as they stand
-// when it begins — and runs an item only if, when its turn comes, it is still equipped on that holder and the holder
-// is still owned: handlers move and destroy pieces mid-walk (突变细胞 transforms its holder and returns the equipment
-// to the hand; normal 博士投影 destroys itself, which splices holder.items), and a live walk skipped the next item or
-// ran it with a holder that was gone. Pieces gained or equipped during the walk wait for the next dispatch.
+// when it begins (every [holder, item] pair, taken before the first item runs) — and runs an item only if, when its
+// turn comes, it is still equipped on that holder and the holder is still owned: handlers move and destroy pieces
+// mid-walk (突变细胞 transforms its holder and returns the equipment to the hand; normal 博士投影 destroys itself, which
+// splices holder.items), and a live walk skipped the next item or ran it with a holder that was gone. Pieces gained or
+// equipped during the walk wait for the next dispatch; an item moved during it runs at most once (on the holder it
+// stood on, if its turn came before the move).
 // Dispatch order per player: global → band → bonds → garrisons (board reading order,
 // then hand) → equipped items → effects (insertion order); onPrice runs the priced chess's own 特质 first (购买价格为N
 // sets the price the discounts and caps of bonds / strategies then act on — user playtest #5). Every call is
@@ -212,15 +214,17 @@ export class EffectDispatcher {
       }
       // 4. garrisons (onPrice: already run as step 0)
       if (hook !== 'onPrice') this._garrisons(ps, hook, ev);
-      // 5. equipped items (not for the item-specific hooks): a snapshot of the owned chess and of each holder's items,
-      // each run only while still equipped on its still-owned holder — handlers move / destroy pieces (header)
+      // 5. equipped items (not for the item-specific hooks): every [holder, item] pair of the owned chess, taken before
+      // the first item runs; each runs only while still equipped on its still-owned holder — handlers move / destroy
+      // pieces (header). Taking the pairs per holder as the walk reached it ran an item equipped meanwhile onto a later
+      // holder, and twice an item moved from a holder already walked to a later one.
       if (hook !== 'onEquip' && hook !== 'onArt' && hook !== 'onDestroy') {
-        for (const holder of ownedChess(ps)) {
-          for (const it of (holder.items || []).slice()) {
-            const key = `item:${itemKey(it.id)}`;
-            const h = reg.get(key);
-            if (h && stillEquipped(ps, holder, it)) this._call(ps, key, h, hook, { kind: 'item', key, piece: it, holder, item: this.m.gd.item(it.id) }, ev);
-          }
+        const pairs = [];
+        for (const holder of ownedChess(ps)) for (const it of holder.items || []) pairs.push([holder, it]);
+        for (const [holder, it] of pairs) {
+          const key = `item:${itemKey(it.id)}`;
+          const h = reg.get(key);
+          if (h && stillEquipped(ps, holder, it)) this._call(ps, key, h, hook, { kind: 'item', key, piece: it, holder, item: this.m.gd.item(it.id) }, ev);
         }
       }
       // 6. persistent effects