Fix TestQueryExecutorConsolidatorWaiterCapReject: set consolidator total waiter count

The test was setting per-query WaiterCount but not the consolidator's
TotalWaiterCount, which is what the actual code checks to decide whether
the cap is exceeded. Also corrected assertions to match the actual code
path (one AddWaiterCounter(-1) call, not two).

Co-Authored-By: Claude <svc-devxp-claude@slack-corp.com>

AI disclosure: Claude Code assisted with development. Every line of code was either written by or carefully reviewed by me :)

Author: bgwines

go/vt/vttablet/tabletserver/query_executor_test.go

@@ -1574,6 +1574,7 @@ func TestQueryExecutorConsolidatorWaiterCapReject(t *testing.T) {
 
 	// Start with waiter count above the cap (2 > 1), so the condition fails
 	fakePendingResult.WaiterCount = 2
+	fakeConsolidator.SetTotalWaiterCount(2)
 
 	fakeConsolidator.CreateReturn = &sync2.FakeConsolidatorCreateReturn{
 		Created:       false, // Simulate identical query already running
@@ -1602,10 +1603,9 @@ func TestQueryExecutorConsolidatorWaiterCapReject(t *testing.T) {
 	// Verify we did NOT broadcast (because we're not the original)
 	require.Equal(t, 0, fakePendingResult.BroadcastCalls)
 
-	// Verify AddWaiterCounter was called: once with 0 (to check count), once with -1 (cleanup)
-	require.Len(t, fakePendingResult.AddWaiterCounterCalls, 2)
-	require.Equal(t, int64(0), fakePendingResult.AddWaiterCounterCalls[0])  // Check current count
-	require.Equal(t, int64(-1), fakePendingResult.AddWaiterCounterCalls[1]) // Decrement
+	// Verify AddWaiterCounter was called once with -1 (cleanup before reject)
+	require.Len(t, fakePendingResult.AddWaiterCounterCalls, 1)
+	require.Equal(t, int64(-1), fakePendingResult.AddWaiterCounterCalls[0])
 
 	// Verify no database query was executed (rejected before fallback)
 	require.Equal(t, 0, db.GetQueryCalledNum(input))