test(mtconnect): de-race the raised-count step-back test

The one-shot /sample failure was enqueued while the FIRST enumeration could still be
before its own one-shot check (the fake releases the call-count waiter first), so
under suite load the failure hit the first stream instead of the raised one
(intermittent). It is now enqueued after the first stream has consumed a chunk.

Refs #743

Claude-Session: https://claude.ai/code/session_01Ub9hGmYLQvn9XETzw7CMsT

tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.MTConnect.Tests/MTConnectSampleLagTests.cs

@@ -130,8 +130,11 @@ public sealed class MTConnectSampleLagTests
         try
         {
             // The NEXT enumeration (the raised one) is refused the way a real Agent refuses a count above its buffer.
+            // Enqueued only once the first stream has consumed a chunk — i.e. is past its own one-shot-failure check —
+            // so the failure can only ever hit the raised stream (enqueueing earlier raced the first enumeration).
+            await client.PumpAsync(Chunk(Primed, 2, last: 900)).WaitAsync(Watchdog, Ct);
             client.SampleFailures.Enqueue(new InvalidDataException("MTConnectError: INVALID_REQUEST count too large"));
-            for (var i = 0; i < 3; i++)
+            for (var i = 1; i < 3; i++)
                 await client.PumpAsync(Chunk(Primed + (2 * i), 2, last: 900)).WaitAsync(Watchdog, Ct);
 
             await client.WaitForSampleCallsAsync(3).WaitAsync(Watchdog, Ct);