* use dnn_fallback flag to delete mkldnn hardcode

* polish code style

* fix protected error

* fix const error

* fix reduce_op fallback

* fix pool_op fallback

* add Set function of dnn_fallback_

* [CodeStyle][py2] remove `paddle.compat`

* remove compat from `paddle.__init__`

* enable_static in sample code

* Revert "enable_static in sample code"

This reverts commit ffccaa633900154ea5f3d056e746aae9a1927399.

* enable_static in sample code

* fix bugs while cond receives dict as input

* add unittest

* change flatten -> _is_sequence_except_dict

* add author

* fix typo, encourges->encourages

* fix geometric doc

* support selected_rows kernel for multiply in dygraph

* delete useless code of slice in fluid

* fix complie bug

* move slice_array from fluid to phi

* fix strided_slice_op_npu

* gene static graph code for exp

* refactor the doc of exp

* fix bug

* fix bug

* update doc of exp

* fix sparse op

* fix log bugs

* more fix

* fix bugs

* update .flake8 config: remove F522, test=document_fix

* test=document_fix

* [CodeStyle][black] update flake8 config

* empty commit, test=document_fix

* [CodeStyle][black] format dy2static unittests

* format some missing files

* update lineno in test_origin_info

* update lineno in test_error

* update lineno

* Fix F522: remove unused arguments

* add redirect_stderr argument in _run_cmd

* delete GetExpectedKernelType mkldnn of mul_grad

* update mkldnn_op_list, remove mul_grad

* delete GetExpectedKernelType mkldnn of mul

* fix macos inconsistent-missing-override warnings

* fix inconsistent-missing-override error in test

* update config

* re-blacken python code

* temporarily disable date and diff_py_file

* skip a format

Chinese and English return values are inconsistent

Fix the bug where the device memory address appears in abs_grad kernel fallback to CPU. test=kunlun (#47186)

* fix nvprof_nvtx_push interface bug