[AP] Add DCU HYTLASS matmul support. (#79083)

* feat(ap): add DCU support for AP backend

* ap get device type api

* ap unified dcu/gpu template

* ap unified dcu/gpu template

* tmp commit

* general ap template

* fix ap cutlass template bug

* autotune for dcu

* fix ap NVGPU autotune bug

* [AP]: merge matmul_variadic_tpl

* [AP]: move cutlass patch file

* update gitignore

* [AP]: cutlass_matmul ap namespace

* [AP]: add ck patch

* add license

* Implement CompileCommand for dcu.

* [AP] Adapt DCU backend from CK to hytlass and fix DTK 25.04.2 build.

* Optimize unittest.

* Remove ck_patch.

* Add hytlass patch.

* Rename namespace from cutlass to cutlass_patch.

* Reorganize codes.

* Add hip backend.

* Set cutlass_dir through environ.

* Fix stream, setting for ap_path and ap_flags, and conversion of int.

* Fix setting of FLAGS_prim_all.

---------

Co-authored-by: _Aln0 <jiangyonglin@baidu.com>
Co-authored-by: _Aln0 <2629184561@qq.com>