* split reduce_kernel

* rm reduce_kernel in cmake

* split reduce_grad kernels

* fix cmake build error

* format code

* fix standalone_executor_test error

* move reduce kernels into one file

* rename reduce_prod to prod

* move reduce sum/mean from math_kernel into reduce_kernel

* rm comment

* migrate matrix_rank to phi

* migrate eigh and matrix_rank to phi

* fix matrix_rank

* optimize code

* move matrix_rank to phi

* add max functor

* migrate matrix_rank to phi

* optimize code