cherry-pick #42750。

QA反馈，#42750 优化后，solov2模型性能可提升6%，故cherry-pick到2.3。因#41096 将linspace python实现从fluid.layers.tensor挪到了paddle.tensor.creation下，该pr不在release/2.3分支中，故将#42750 中python修改同步到fluid.layers.tensor.linspace中。

[cherry pick] Support optional residual add in fused ops and slice large tensor for cudnn_softmax (#43719)

 [cherry pick] Support optional residual add in fused ops and slice large tensor for cudnn_softmax

cherry-pick #43635 #43681 #43474

* cherry pick from #43397

* fix code

* fix bug of strided_slice (#43388)

* fix stride_slice bug

* fix bug

* fix bug of infer shape for slice (#43443)

* [EinsumOp] Polish forward logic and backward logic for optimize (#42603)

* change logic for optimize

* modifty

* merge

* change einsum_v2 as default and add new flags: FLAG_einsum_opt=1|0 (#43010)

* [EinsumOp] Make EinsumOp support bfloat16. (#43085)

* change einsum_v2 as default and add new flags: FLAG_einsum_opt=1|0

* make EInsumOP support bf16

* add unittest for BF16

* add condition for test_BF16

* fix bugs

* fix

* change the backward api to fit einsum op

Reduce amax/amin frobenius_norm_kerne原始实现为Eigen实现，文件编译时间较长，因此本PR将其替换为KP实现
删除DefaultElementwiseOperator中重复功能支持，减少elementwise_double_grad OP编译时间

Delete ElementwiseKernel in BroadcastKernel
减少所有Broadcast中重复功能调用，同时减少编译时间和问题体积

删除Broadcast function中rank例化以及Elementwise调用，降低编译时间。
从develop分支中的#42645 PR修改而来，由于develop分支与release分支相差较大，无法实现cherry-pick，因此针对release2.3重新提交PR.
Broadcast中关于rank的例化会导致底层模板展开较多，造成reduce_sum_grad_kernel.cu.o文件体积过大，修改后可以降低.o体积及编译时间

* Fix the race condition in cumsum operator

* Optimize cumsum operator
Co-authored-by: NLeo Chen <39020268+leo0519@users.noreply.github.com>

* fix bug of batch_norm_grad kernel with fp16

* format code

* Extend python einsum interface to make einsum_v2 support multi-operands and switch it to default.

* add opt_einsum dependence

* add yaml and support eager model

* fix by code review

* opt attr eaque perf

* opt attr select code

* fix one hot infermeta

* polish get attr impl

* fix tests failed

* add testcases

* full api fix

* when out is None, go old dygraph mode

* by static check

* first version: support 2-inputs forwards. TODO: 1. backward  2. BroadCast  3. MultiVariable

* time out -> 120

set device id of Place() to get GPUContext needed by LimitGridDim in ElemwiseGradBroadcast (PaddlePaddle#42320) (#42332)

* Optimize performance of dygraph (v4)  (#42196)

* optimize performance of dygraph

* optimize performance of dygraph and elementwise_add

* optimize the trace op

* fix bug

* fix bug

* fix unittest bug

* fix code format

* fix cherry-pick problem

* Optimize the performanece of sum api (#42231)

* optimize the performanece of sum api

* optimize IsDenseTensorInput

* remove debug log

* Add move construct for KernelSignature (#42253)

* add move construct for KernelSignature

* add noexcept

* fix cherry-pick problem

* Add paddle::variant and replace paddle::any (#42139)

* add variant and replace any

* split attribute

* Optimize dygraph GetExpectedKernelType perf (#42154)

* opt dygraph scheduling

* revert part impl

* fix variant compile error (#42203)

* replace any by variant in infermeta (#42181)

[Cherry-Pick][Performance]Remove CudaStreamSychornize in ClipGradByGlobalNorm and fix shape op (#42170)

* [Performance]Set ShapeKernel with ALL_BACKEND and ALL_LAYOUT (#42138)

* [Performance]Set ShapeKernel with ALL_BACKEND and ALL_LAYOUT

* [Performance]Set ShapeKernel with ALL_BACKEND and ALL_LAYOUT

* [Performance]Remove CudaStreamSychornize in ClipGradByGlobalNorm (#42132)

* [Phi] Support setting size of vector<Tensor> for out in yaml (#41576)

* support setting vector out size in yaml

* support setting size of vector<tensor> for out in yaml

* resolve conflict
Co-authored-by: Nzyfncg <zhangyunfei07@baidu.com>

* cherry-pick python/paddle/utils/code_gen/backward.yaml

* remove unsupported yaml
Co-authored-by: NZhanlue Yang <jim19930609@gmail.com>

* add rsqrt yaml and unittest (#41443)

* Add expand equal all yaml (#41540)

* add expand, poisson

* add poison grad

* add expand equal_all poisson triangular solve yaml
Co-authored-by: Nhong <43953930+phlrain@users.noreply.github.com>

Cherry-pick #40338 #41741 #41313

cherry-pick the PR#41586 to realese/2.3

* add eids result for graph_sample_neighbors

* fix bug

* move fisher_yates sample to warp

* add cpu eid output

* delete comment

* delete comment

* change nullptr placeholder

* optimize sample kernel

* fix mutable_data

* split reduce_kernel

* rm reduce_kernel in cmake

* split reduce_grad kernels

* fix cmake build error

* format code

* fix standalone_executor_test error

[DoubleGrad] Enabled double grad test cases in eager_mode for test_imperative_double_grad (#41451) (#41893)

* [DoubleGrad] Enabled double grad test cases in eager_mode for test_imperative_double_grad

* Fixed elementwise issue

* Addressed CI failures

Add paddle.sparse and three Sparse API (#41276)
Add Sparse API to_dense, to_sparse_coo and values (#41394)

* add multi_dot,maxout,multiplex yaml

* add code converage

* [Yaml]add exp yaml (#41217)

* add exp yaml

* add exp api in test case

* add determinant yaml

* fix exp op unittest

* change test class name

* modify api name

* compacted with raw api

* fix det api

* add python_api

* add test eager for determinant op

* [Yaml] Add assign yaml (#41428)

* add assign yaml

* add assign api

* add assign backward api

* add assign

* add assign yaml

* add assign

* assign yaml

* add assign raw kernel and use assign_raw in yaml

* merge develop branch

* add missing python_api

* exchange assign and assign_raw kernel name (#41625)

* exchange assign and assign_raw kernel name

* fix register error

* [Yaml]add gaussian_random yaml and test case (#41312)

* add guassian random yaml

* add gaussian_random yaml and test case

* fix error modify of full yaml

* import in_dygraph_mode

* import _in_legacy_dygraph

* add place arg in api

* import __current_expected_place

* fix test_egr_python_api failed case

* add test case

* add cast for NormalInitializer

* fix test error

* fix test error

* rm unsed check code

* fix test error in test_initializer_nn

* modify by review

* [Phi]fix split error when sections has 0 size and add test case (#41708)

* fix split error when sections has 0 size and add test case

* fix test case

* [Eager] Remove elementwise add in conv (#41515)

* remove elementwise add in conv

* use reshape

* fix warpctc grad kernel dep eror (#41598)

add a inner loop for index_select_grad_init() in index_select op when dealing with large-shape data (PaddlePaddle#41563) (#41669)

* Revert "[Phi] Migrate Adam and AdamW into Phi (#40351)"

This reverts commit 56cd3407.

* add infermeta

* fix search sort bug (#41664)

* fix depthwise dnn bug (#41666)

[Cherry-Pick]Add hard_swish/kron/linspace/logit/graph_send_recv/multi_dot/maxout/multiplex op yaml file  (#41566)

* [Phi]Add graph_send_recv yaml file (#41206)

* add graph_send_recv yaml

* deal with confict

* fix compile bugs

* cherry-pick pr 41298

* cherry-pick pr41550

* fix compile bugs

[Cherry-pick] Add truncated_normal/unique/swish/unbind yaml and polish Getting tensor place impl (#41539)

* [Phi] Polish truncated normal kernel and add yaml (#41280)

* polish truncated normal kernel

* add yaml

* add truncated normal kernel and add yaml

* polish unittests and yaml

* import dygraph mehtod

* add unique yaml and final state api (#41460)

* fix get tensor backend set bug (#41478)

* [Phi] Add unbind yaml and final state api (#41277)

* add unbind yaml

* fix unittest

* [Phi] Add swish yaml and final state api (#41479)

* add swish yaml and final state api

* skip mkldnn test

* fix grad mkldnn test

* add cherry-pick lost code