[ARC-WSL2][pytorch 2.13] torch.xpu.is_available() get Segmentation fault (core dumped)

### 🐛 Describe the bug

" torch.xpu.is_available()" get Segmentation fault (core dumped) on ARC-WSL with pytorch 2.13
```
Python 3.10.20 | packaged by conda-forge | (main, Jun 11 2026, 03:31:56) [GCC 14.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.version.git_version
'a3aa2c6ee42a053b5f761f4ff1f47974c39714fb'
>>> torch.xpu.is_available()
Segmentation fault (core dumped)
```

### Versions

WSL Driver:
Driver UUID 32342e35-322e-3332-3232-340000000000
Driver Version 24.52.32224

Windows Driver: 32.0.101.8331

Start by reproducing the reported Python snippet with torch.xpu.is_available() under the stated PyTorch, Python, WSL2, and driver versions. The entry point is torch.xpu.is_available(); done means it no longer causes a segmentation fault and returns a valid availability result, with the behavior checked against the reported environment.

Written by the indexing model from the issue text.