* delete fill xpu op in fluid

* delete fill_constant header, test=kunlun

* fix npu header, test=kunlun

* remove cudaDeviceContext

* remove more template

* fix rocm compile

* remove alias name CUDADeviceContext

* fix compile

* fix tests

* revert changes

* remove is_init_py of RandomGenerator, and use Global Generator if not OP seed

* fix comment

* add gaussian random

* add full

* refine reduce

* refine code

* refine gaussian_random unittest

* add unittest for fill_any_like fill_constant

* clean distribution_helper, index_impl, aligned_vector code in fluid

* fix conflicts

[Phi] move gaussian_random kernel

* Modified RandomKernel with Kernel Primitive API

* update pten.h to phi.h

* update

* update fullKernel

* change CUDA implementaion of uniform/gaussian OP

* fix unittest

* add pten::Place data structure.

* update ci problem

* fix ci problem

* update

* using platform::Place=pten::Place

* remove BOOST_GET_CONST for CPUPlace and GPUPlace

* compile pass 25%.

* compile pass 45%

* compile pass 60%

* remove boost_get for xpu npu mlu and ipu

* compile pass on cpu and gpu.

* fix compile problem

* fix compile error.

* update

* fix ci problem

* update

* ci approve

* fix ci problem

* fix ci eager test problem

* remove BOOST_GET_CONST

* fix npu compile

* remove thrust includes, test=develop

* fix compilation error, test=develop

* fix compilation of truncated_gaussian_random_op, test=develop

This reverts commit bf3c3496.

* "cherry picked operators changes"

* "remove duplicated code"

* "add constant setter"

* "add get expected kernel"

* "fix ci"

* "add fill constant"

* Generating random numbers with given batch size

uniform_random_batch_size_like_op
gaussian_random_batch_size_like_op

* More comments about random seed.

* Move test_*_random_batch_size_like_op to unittests

There are mainly following fixes:

- take `DeviceContext` as the template parameter of math functors and OpKernel instead of `Place`
- remove `eigen_device` interface in base class  `DeviceContext`
- remove `GetEigenDevice` interface in `ExecutionContext` and base class `DeviceContext`
- remove unused `platform::EigenDeviceConverter`
- rename `REGISTER_OP_GPU_KERNEL` to `REGISTER_OP_CUDA_KERNEL`
- rename `USE_GPU_ONLY_OP` to `USE_CUDA_ONLY_OP`

Fix #3902

* Should always use `dynload::` for cuda function.
* Fix cublas.h without DSO load.