perf: allocate deck pull list in parallel instead of serially

Applying a pull list awaited one AllocateDeckCardItem mutation per physical
copy in a sequential loop, so a 60-card pull meant 60 serial round trips.

Run the entries through Promise.all instead. Each entry targets a distinct
collection item (within a deck an item maps to a single deck card), so entries
touch disjoint rows and can't race; copies of the same item within an entry
stay serial to avoid racing on that item's availability. Wall-clock drops from
the sum of all copies to the longest single entry.

Addresses #104; collapsing the bulk update/delete Promise.all loops needs new
server-side bulk mutations, tracked in #129.

Author: cfbender

assets/react/src/pages/decks/detail-page.tsx

@@ -651,8 +651,13 @@ export function DeckDetailPage({
     isPending: isAllocateDeckPullListPending,
     mutate: (entries: DeckPullListEntry[]) => {
       setIsAllocateDeckPullListPending(true)
-      void (async () => {
-        for (const entry of entries) {
+      // Allocate entries in parallel instead of one serial round trip per copy.
+      // Each entry targets a distinct collection item (within a deck an item maps
+      // to a single deck card), so entries touch disjoint rows and can't race.
+      // Copies of the same item within an entry stay serial to avoid racing on
+      // that item's availability.
+      void Promise.all(
+        entries.map(async (entry) => {
           for (let copy = 0; copy < entry.quantity; copy += 1) {
             await client.mutate({
               mutation: AllocateDeckCardItemDocument,
@@ -662,8 +667,8 @@ export function DeckDetailPage({
               },
             })
           }
-        }
-      })()
+        }),
+      )
         .then(() => {
           const allocatedCount = entries.reduce((total, entry) => total + entry.quantity, 0)