* Revert "Revert "【Hackathon No.67】remove operator.h in blas.h (#50989)" (#51467)"

This reverts commit b9d91531.

* remove cout

* add header

* fix missing header

* fix refer fluid error

* fix missing header

* 更新 repeat_interleave_grad_kernel_impl.h

Change to phi style datatype.

* 更新 repeat_interleave_grad_kernel_impl.h

Fix missing header

* datatype fluid -> phi

* paddle::experimental -> phi

* fix reference error

* fix reference error

* fix reference error

* fix errors

* fix missing FLAGS

* fix missing headers

* fix missing headers

* fix missing headers

* fix missing headers

* fix missing header

* fix missing header

* fix errors

* Delete randperm raw op

* fix

* test_get_kernel

* add invoke signature

* change reduce_max

* change frobenius_norm

* reset reduce_max according to composite and change reduce_all

* fix the bug when Scalar(*)

* fix 'scalar when support_tensor'

* change code according to review

* change 'keep_signature' to 'manual_signature' and add some erro info

* remove parallel_executor related unit tests

* fix CI

* support to_static for SpectralNorm

* support amp logic for layer_norm and softmax

* fix layer_norm amp

* fix layernorm api and dropout fp16

* fix layernorm api and dropout fp16

* fix bn, ln dtype in float16

* fix dropout fp16

* fix comment

* remove with_data_parallel in example code

* revert python/paddle/fluid/data_feeder.py

* fix static.nn.fc api

* support gather test on prim and cinn

* reset timeout for gather

* add pow composite rule

* fix test

* fix unit test

* update test

* fix test

* update

* add fp16 test for divide, matmul, pnorm

* add cumsum fp16 unittest

* fix threshold

* revert cumsum

* fix code-style

* fix according to review

* fix kernel not found

* Delete hardswish_raw op

* fix ut

* [CustomDevice] fix SyncDefaultStream for process_group_custom

* update

* refine _found_inf

* modify_yaml

* delete default param

* add output for matmul_double_grad

* add split_with_num composite rule

* add split_with_num composite rule

* add split composite rule

* update

* update test

* update test

* delete split_with_num_grad

fix abi

fix tab

* first commit

* fix code bugs in for_loop

* fix bugs in cuLoadAddStridedInputs.

* optimization for LayerNormBackwardComputeGradInput

* add unitest for validating the optimization

* fix windows ci error

* cuda graph support multi-stream for new executor

* fix windows compile error

* delete create_cuda_graph_stream

* wisemax fp16 support

* add bf16 support 4 elementwise_max

* append broadcast 4 op 4 fp16 / bf16

* fix elewise_max ut bf16 numeric delta

* append fp/bf16 uts

* add fp/bf16 uts

* change bf16 uts delta

* fix some issue

* add prim 4 fp16