Fix build errors of GridSample and test failures in test_attention_fusion.py (#27642)

# Description

This PR addresses a build error and subsequent test failures related to
recent changes in GridSample and the transformer optimizer. Related PRs:
#27201, #27556.

## Changes

### 1. Fix GridSample Build Error
- Removed an unused local variable `mode_str` in
`onnxruntime/core/providers/cuda/tensor/grid_sample.cc` that was causing
a warning (treated as error) about shadowing a member variable.
- Ref:
[`grid_sample.cc`](https://github.com/microsoft/onnxruntime/blob/c979a2407f/onnxruntime/core/providers/cuda/tensor/grid_sample.cc#L54)

### 2. Update GridSample Tests
- Updated
`onnxruntime/test/providers/cpu/tensor/grid_sample_test_custom.inc` to
use default execution providers in `RunTests` instead of a hardcoded
opset version, ensuring compatibility across different environments.

### 3. Revert Transformer Fusion Fallback
- Reverted a recent change in
`onnxruntime/python/tools/transformers/fusion_skiplayernorm.py` that
enabled a fallback for `SkipLayerNormalization` fusion when symbolic
shape inference fails.
- This revert was necessary to avoid regressions in GPT-2 tests where
model definitions contain typos that intentionally (or coincidentally)
break shape inference.
- Ref:
[`fusion_skiplayernorm.py`](https://github.com/microsoft/onnxruntime/blob/c979a2407f/onnxruntime/python/tools/transformers/fusion_skiplayernorm.py#L113)

### 4. Restore Transformer Test Parity
- Updated
`onnxruntime/test/python/transformers/test_attention_fusion.py`
specifically `test_qwen3_normalization_fusion` to match the expected
node counts after reverting the fusion fallback.
- Ref:
[`test_attention_fusion.py`](https://github.com/microsoft/onnxruntime/blob/c979a2407f/onnxruntime/test/python/transformers/test_attention_fusion.py#L398)

## Verification

- `build_cuda.sh` completed successfully.
- `onnxruntime/test/python/transformers/test_attention_fusion.py` passes
with "OK".
- `lintrunner -a` reports no issues.

Author: tianleiwu

onnxruntime/core/providers/cuda/tensor/grid_sample.cc

@@ -51,7 +51,6 @@ template <typename T, bool IsNHWC>
 GridSample<T, IsNHWC>::GridSample(const OpKernelInfo& info) : CudaKernel(info) {
   opset_start_version_ = info.node().SinceVersion();
 
-  std::string mode_str = info.GetAttrOrDefault<std::string>("mode", "bilinear");
   std::string padding_mode_str = info.GetAttrOrDefault<std::string>("padding_mode", "zeros");
   align_corners_ = static_cast<bool>(info.GetAttrOrDefault<int64_t>("align_corners", 0));
 

onnxruntime/python/tools/transformers/fusion_skiplayernorm.py

@@ -110,9 +110,8 @@ def fuse(self, node, input_name_to_nodes, output_name_to_node):
                     )
                     return
             else:
-                # Shape inference failed. Use default skip_index=1 (no broadcasting) since both
-                # Add inputs have already been verified as non-initializer dynamic tensors above.
-                logger.debug("symbolic shape inference failed, using default skip_index for SkipLayerNormalization")
+                logger.debug("skip SkipLayerNormalization fusion since symbolic shape inference failed")
+                return
 
         gather_path = self.model.match_parent_path(add, ["Gather"], [None])
         if gather_path is not None and self.model.find_graph_input(gather_path[0].input[1]) is None:

onnxruntime/test/providers/cpu/tensor/grid_sample_test_custom.inc

@@ -37,7 +37,7 @@ TYPED_TEST(GridSampleCustomTest, test_grid_sample_20_4D_linear_zeros_mixed_bound
   test.AddAttribute("padding_mode", padding_mode);
   test.AddAttribute("align_corners", align_corners);
   test.AddOutput<TypeParam>("Y", Y_shape, Y_data);
-  RunTests(test, GetExecutionProviders(20));
+  RunTests(test, GetExecutionProviders());
 }
 
 TYPED_TEST(GridSampleCustomTest, test_grid_sample_20_4D_linear_zeros_mixed_bounds_left_top) {
@@ -69,6 +69,5 @@ TYPED_TEST(GridSampleCustomTest, test_grid_sample_20_4D_linear_zeros_mixed_bound
   test.AddAttribute("padding_mode", padding_mode);
   test.AddAttribute("align_corners", align_corners);
   test.AddOutput<TypeParam>("Y", Y_shape, Y_data);
-  RunTests(test, GetExecutionProviders(20));
+  RunTests(test, GetExecutionProviders());
 }
-

onnxruntime/test/python/transformers/test_attention_fusion.py

@@ -395,17 +395,17 @@ def test_qwen3_normalization_fusion(self):
         ssln_count = sum(1 for n in nodes if n.op_type == "SkipSimplifiedLayerNormalization")
 
         # 4 RMSNorm patterns: pre-attn, Q-norm, K-norm, post-attn.
-        # Post-attn RMSNorm has an Add parent (residual) → fused as SkipSimplifiedLayerNormalization.
-        # Remaining 3 stay as SimplifiedLayerNormalization.
+        # Fallback for SkipLayerNormalization is disabled, so post-attn RMSNorm does not fuse.
+        # All 4 stay as SimplifiedLayerNormalization.
         self.assertEqual(
             sln_count,
-            3,
-            f"Expected 3 SimplifiedLayerNormalization (pre-attn + Q-norm + K-norm), got {sln_count}",
+            4,
+            f"Expected 4 SimplifiedLayerNormalization (pre-attn + Q-norm + K-norm + post-attn), got {sln_count}",
         )
         self.assertEqual(
             ssln_count,
-            1,
-            f"Expected 1 SkipSimplifiedLayerNormalization (residual + post-attn RMSNorm), got {ssln_count}",
+            0,
+            f"Expected 0 SkipSimplifiedLayerNormalization (residual + post-attn RMSNorm failed to fuse), got {ssln_count}",
         )