Implementation of matmul for complex datatypes.

[PawelSwider2000]

Implementation of kernels for complex datatype support for 4 ops: mm, bmm,addmm, baddbmm using OneMKL.

Current implementation of this ops for XPU is in pytorch/aten/src/ATen/native/mkldnn/xpu/Blas.cpp. Since OneMKL is a torch-xpu-ops dependency and is available only with USE_ONEMKL_XPU=ON (which is a default value). Implementation needs to be in torch-ops-xpu and kernels and TORCH_LIBRARY_IMPL are in ifdef macro to avoid complication error when OneMKL is not supported. Newly declared op will be called from existing torch implementation using c10::Dispatcher.

This is part of: #1853

[Copilot]

Pull Request Overview

This PR implements OneMKL-based kernels for complex datatype support in matrix multiplication operations (mm, bmm, addmm, baddbmm) on XPU devices. The implementation provides optimized BLAS operations for complex numbers using OneMKL library integration.

Key changes include:

• Addition of OneMKL-based complex matrix multiplication kernels
• Implementation of four core matrix operations with complex number support
• Conditional compilation support for OneMKL availability

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[PawelSwider2000]

@CuiYifeng @kbinias Please review

[PawelSwider2000]

Follow up change with tests: #1993

[Copilot]

Pull Request Overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 3 comments.

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}