[Eager, Performance optimization] math op sink to Cpp level ( + and - operator as an example ) (#45811)

* [Eager] math op sink to Cpp level

* fix ci errors

* draft version

* draft version

* draft version

* support + and - operator under cpp directly

* add static test

* polish code

* promote types or unify right type to left

* recover static test case

* polish code and fix some ci errors

* support complex and polish code

* fix conflicts

* fix windows ci errors

* fix windows-inference-ci errors

* polish and fix tests

* fix test case

* polish code

* polish code

* polish code and fix code-format

* polish code

* polish code

* polish code

* polish code

* enable eager mode on xpu, test=kunlun

* add numpy support to xpu

* fix tensor using error

* fix  error, test=kunlun

* fix failed tests, test=kunlun

* support rsqrt_p

* refine code and ut

* add_prim_rsqrt

* fix ut

* open the clip_extra flag in paddle.static.save_inference_model, test=allcase

* set the defalut value of clip_extra in TracedLayer from False to True, test=allcase

*  Disable TF32 to solve accuracy for test_trt_conv_pass and test_trt_deformable_conv in A10 GPU.

* remove `yapf: disable` from math.py

* disable test case only

* remove useless disable

* remove useless disable

* revert no_grad_set_white_list

* add phi reduce_sum test=kunlun

* add fhi reduce_sum test=kunlun

* add fhi reduce_sum test=kunlun

* 增加ks-test；注册kl_divergence

* bug fix

* bug fix data shape

* 按要求修改

* sample code fix

* 按要求修改

* bugfix

* bugfix

* bugfix

* bugfix

* bugfix

* fix sample func

* fix sample func

* 按要求修改；sample函数bugfix

* 按要求修改；sample函数bugfix

* bugfix

* 补充英文文档公式

* 英文文档bugfix

* bugfix

* bugfix

* 英文文档修复

* 去除name参数

* [BugFix]Fix reduce_mean/min/sum/prod, cumsum grad_op infershape bug

* fix typo

* fix typo

* add numpy api err msg

* fix bug

* fix unittest

* add set_state_dict err

* rewrite numpy_api_check

* add define

* change err msg

* fix test

* move import statement

* fix sub sign reverse for mkldnn

* refactor code as comment

* remove useless

* format code

* fix beta bug in fill_constant_batch_size_like

* place

* add test

* fix spell and test

* fix anno

* code style

* [Check]Enhance pooling output_size type check

* add unittest

* use re replace judge by case

* simplify re

* for add_bias

* residual_no_bias

* comments

* more ut

* fix input

* Remove trt_reshape2_matmul_fuse_pass

* [Eager] polish clone interface

* rm clone in python, add clone in eager_method.cc

uniform logger manager in FleetAPI.
hidde API under distributed/utils which users don't need.

* Support matmul_v2 in PaddleTensorRT

* move_merge_selected_rows

* update code