fix(table): seed dispatch overlay on Run; surface batch-enqueue failures as error

- useRunColumn.onSuccess invalidates activeDispatches so the resolveCellExec
  queued overlay populates immediately for ahead-of-cursor rows instead of
  waiting for the first dispatch SSE. Rows cache stays owned by SSE.
- On batchEnqueueAndWait failure, dispatcherStep flips orphan pre-stamps to a
  terminal error state (+ cell SSE) instead of deleting them, so the dropped
  window is visible (Error pill) rather than silently empty and re-runs on the
  next manual run.

Author: TheodoreSpeaks

apps/sim/hooks/queries/tables.ts

@@ -1378,14 +1378,8 @@ export function useRunColumn({ workspaceId, tableId }: RowMutationContext) {
       if (context?.snapshots) restoreCachedWorkflowCells(queryClient, context.snapshots)
     },
     onSuccess: () => {
-      // Seed the active-dispatch overlay from the server. `runWorkflowColumn`
-      // inserts the dispatch row synchronously before responding, so the
-      // refetch picks it up immediately — this drives `resolveCellExec`'s
-      // queued overlay for rows ahead of the cursor (scrolled-in / refetched
-      // rows that have no real exec yet) without waiting for the first
-      // `kind: 'dispatch'` SSE. Targeted at activeDispatches only — NOT the
-      // rows cache, which `useTableEventStream` owns (a rows refetch would
-      // race its incremental patches).
+      // Seed the active-dispatch overlay immediately (insertDispatch ran
+      // server-side before responding); rows cache stays owned by SSE.
       void queryClient.invalidateQueries({ queryKey: tableKeys.activeDispatches(tableId) })
     },
   })

apps/sim/lib/table/dispatcher.ts

@@ -399,15 +399,8 @@ export async function dispatcherStep(dispatchId: string): Promise<DispatcherStep
       logger.error(`[${dispatchId}] batch dispatch failed`, {
         error: toError(err).message,
       })
-      // The enqueue failed (e.g. trigger.dev API down) — these cells never
-      // ran. The cursor still advances past this window below, so they won't
-      // be retried in this dispatch. Flip the orphan pre-stamps to a terminal
-      // `error` state (with an SSE event) instead of deleting them: the
-      // failure stays visible to the user (Error pill, not a silently-empty
-      // cell), the cells aren't stuck `pending` forever, and `error` cells
-      // re-run on the next manual run (classifyEligibility treats error as
-      // eligible for manual / incomplete). Guarded on `pending + null
-      // executionId` so we only touch our own un-claimed pre-stamps.
+      // Cursor advances past this window, so flip the un-claimed pre-stamps to
+      // terminal `error` (+ SSE) — visible, not stuck pending, re-runnable.
       const failedAt = new Date()
       await Promise.allSettled(
         pendingRuns.map(async (p) => {