DirectML support on Windows

[divideconcept]

🚀 The feature, motivation and pitch

Executorch is not officially supporting Vulkan for Windows.
CUDA support has been recently added, but only works with NVIDIA GPUs.
DirectML support would provide GPU acceleration for all hardware manufacturers on Windows : NVIDIA, AMD, Intel, and Qualcomm.
It works quite well when used with ONNX Runtime.

Alternatives

No response

Additional context

No response

RFC (Optional)

No response

[JacobSzwejbka]

If anyone wants to try adding some DirectML support we are happy to have contributions. Expansion of Vulkan support is being discussed as well.