Enable GGUF support gating for XPU in quantization compatibility checks

[Copilot]

This update addresses XPU GGUF enablement by removing a CUDA/ROCm-only assumption in quantization support detection. The GGUF path is now recognized as supported on XPU, so GGUF coverage is no longer blocked by platform gating logic.

• Problem addressed

  • Quantization support checks treated non-CUDA/ROCm platforms as unsupported by default.
  • On XPU, this incorrectly rejected gguf and skipped GGUF-specific validation paths.

• What changed

  • Updated tests/quantization/utils.py::is_quant_method_supported to:
    • allow gguf on current_platform.is_xpu()
    • handle None device capability (XPU behavior) without failing, while still restricting non-GGUF methods
  • Preserved existing behavior for CUDA/ROCm and capability-based checks.

• Focused coverage added

  • New tests in tests/quantization/test_utils.py verify:
    • XPU + gguf => supported
    • XPU + non-gguf (e.g. fp8) => unsupported

if capability is None:
    return current_platform.is_xpu() and quant_method == "gguf"

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.