[DML EP] Add intermediate tensor dumping for DML (#22246)

### Description
<!-- Describe your changes. -->



### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->

onnxruntime/core/framework/debug_node_inputs_outputs_utils.cc

@@ -333,7 +333,7 @@ void DumpTensor(
   } else {
     std::cout << tensor_location << "\n";
 
-#if defined(USE_CUDA) || defined(USE_ROCM)
+#if defined(USE_CUDA) || defined(USE_ROCM) || defined(USE_DML)
     const auto data_type = tensor.DataType();
     // Dumping GPU only when cuda is enabled.
     if (tensor_location.device.Type() == OrtDevice::GPU) {