* support pure bfloat16

* support bf16 linear

* update PR to pass CI

* tiny fix where_grad_kernel.cu

* add bfloat16 to selu_grad to pass CI

* fix selu grad compilation error

resort .cu headers, set clang-format not sort include block and consider .cu as main source file (#43633)

* rename scalar_array to int_array

* update cmake

* fix conflict

* remove useless log

* Add stft_op.

* Add stft_grad_op.

* Add stft_op unittest.

* [DLTP-45176] Add complex compatibility in static mode for stft api.

* [DLTP-45176] Add complex compatibility in static mode for stft api.

* Add doc.

* Update unitests of stft op.

* Update spectral helper.

* fix coding style.

* add pad3d kernel into phi

* add pad3d infermeta

* fix build error

* remove raw pad3d infershape function