XPU does not support NestedTensor for SDPA operations.

### 🐛 Describe the bug

Cases:
op_ut,third_party.torch-xpu-ops.test.xpu.functorch.test_vmap_xpu.TestVmapBatchedGradientXPU,test_randomness_backend3_randomness_different_xpu
op_ut,third_party.torch-xpu-ops.test.xpu.functorch.test_vmap_xpu.TestVmapBatchedGradientXPU,test_randomness_backend3_randomness_error_xpu
op_ut,third_party.torch-xpu-ops.test.xpu.functorch.test_vmap_xpu.TestVmapBatchedGradientXPU,test_randomness_backend3_randomness_same_xpu
op_ut,third_party.torch-xpu-ops.test.xpu.functorch.test_vmap_xpu.TestVmapBatchedGradientXPU,test_sdpa_backend3_xpu
op_ut,third_party.torch-xpu-ops.test.xpu.test_nestedtensor_xpu.TestNestedTensorSubclassXPU,test_sdpa_with_packed_in_proj_xpu_bfloat16

The feature block following tests.

test/test_nestedtensor.py | TestNestedTensorSubclassXPU | test_dummy_mha_with_nt_use_legacy_api_False_xpu
-- | -- | --
test/test_nestedtensor.py | TestNestedTensorSubclassXPU | test_dummy_mha_with_nt_use_legacy_api_True_xpu

```
===================================================================== FAILURES =====================================================================
___________________________________ TestNestedTensorSubclassXPU.test_dummy_mha_with_nt_use_legacy_api_False_xpu ____________________________________ Traceback (most recent call last):
File "/home/daisyden/upstream/inductor4/third_party/torch-xpu-ops/test/xpu/test_nestedtensor_xpu.py", line 7490, in test_dummy_mha_with_nt
attn_output_eager, _, _ = m_eager(query, value, offsets)
File "/home/daisyden/miniforge3/envs/inductor4/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1779, in _wrapped_call_impl return self._call_impl(*args, **kwargs)
File "/home/daisyden/miniforge3/envs/inductor4/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1790, in _call_impl
return forward_call(*args, **kwargs)
File "/home/daisyden/upstream/inductor4/third_party/torch-xpu-ops/test/xpu/test_nestedtensor_xpu.py", line 7449, in forward
attn_output = torch.nn.functional.scaled_dot_product_attention(
File "/home/daisyden/miniforge3/envs/inductor4/lib/python3.10/site-packages/torch/nested/_internal/nested_tensor.py", line 389, in __torch_function__
return jagged_torch_function(func, *args, **kwargs)
File "/home/daisyden/miniforge3/envs/inductor4/lib/python3.10/site-packages/torch/nested/_internal/ops.py", line 390, in jagged_torch_function
return jagged_scaled_dot_product_attention(*args, **kwargs)
File "/home/daisyden/miniforge3/envs/inductor4/lib/python3.10/site-packages/torch/nested/_internal/sdpa.py", line 934, in jagged_scaled_dot_product_attention
raise RuntimeError(
RuntimeError: No viable backend for scaled_dot_product_attention was found.

To execute this test, run the following from the base repo dir:
python test/xpu/test_nestedtensor_xpu.py TestNestedTensorSubclassXPU.test_dummy_mha_with_nt_use_legacy_api_False_xpu

```

### Versions

main