* 1, remove layernorm dynamic fp16; 2, let reshape out in dynamic shape (#33535)

* add win_capi_tar in paddle_build.bat

* tar capi lib for publish

* add in gen_fluid_lib func

* Fix gather infer shape using axis (#33413)

* fix gather shape bug

* fix None

* fix topo

* Fix hang of hybrid parallel in new_group  (#33141)

* fix hang of hybrid parallel

* fix new_group for hang problem

* fix hang

[cherry-pick] Polish code for setitem/getitem and support index for list/Tensor/None/Ellipsis/bool (#33528)

* [cherry-pick 2.1] Polish code for _getitem_impl_ (#32868)

* [cherry-pick] Polish code for setitem and getitem (#32911)

* [slice getitem] Support getitem idx is Tensor or List (#33000)

* [getitem] Support index is None for getitem in static mode (#33001)

* [Static getitem] Support static Variable getitem for Ellipsis index (#32876)

* [static getitem]Support index is list bool for getitem in static mode (#33298)

* add warning log for DataLoader output format imcompatible upgrade. test=develop

* add unittest. test=develop

* fix ci converage. test=develop

* fix ci coverage. test=develop

修复pylayer 返回to_tensor时触发段错误的bug。
原因：

如果在Python端修改了stop_gradient属性，c++ 端InnerSetOverridedStopGradient 无法修改stop_gradient属性，在c++端调用SetOverridedStopGradient修改stop_gradient属性。
to_tensor产生的tensor的grad var的DataType为默认值（-1），在backward的过程中grad var的DataType不能为默认值（-1），因此在调用ForwardDataType设置grad var的DataType。

原始PR：#33303

* Eliminate numerical differences of LayerNorm; fix LayerNorm Nan Bug while large data input

* fix bug while large shape of data input

* update 2.0 public api in all left files

* reverse device.py all list;
fix some flake8 errors

* update 2.0 public api in vision

* fix some flake8 errors

* add unit8 for concat (#32850)

* add bool type for tril api (#33402)

Support diff dataset tensor place in single process dataloader

cherry-pick of #33470

 [cherry-pick] use the required instruction to determine if the environment fits the sample code's required.  (#32766) (#33451)

1 put a instruction # required: gpu(for example) in the sample code
2 this piece of code will only run in the GPU-equipped CI pipelines, and be omitted in other pipelines.
3 the CI pipelines can specify its capacity by shell environment variable SAMPLE_CODE_TEST_CAPACITY

2.1 文档改版方案

see #32766 for more infomation

使用op benchmark时发现，当输入数据量小于某个值时，python 端 log_softmax 接口的输入值经过计算过后 会被改变为nan。输出正常。

cherry-pick自 #32937

 fuse L2Decay and momentum when param.regularizer is set

cherry-pick #32845

* add random state generate in DataLoader worker. test=develop

* fix license and __all__. test=develop

* fix unittest. test=develop

* fix retry in download when exception occurs

* add test_retry_exception

* update, test=develop

* fix: paddle.static.default_main_program. test=develop

* add normalize_program to __all__. test=develop

* fix conv2d_transpose trt bugs (#33242)

* fix jetson arch when compiling with single arch (#33269)

* Fix two english api documents, transpose and strided_slice

* OP:strided_slice_op supports bool type inputs

* Fix comments in framework (#32861)

* Fix comments in framework

* Update framework.py

* fix code style
Co-authored-by: NWenyu <wenyu.lyu@gmail.com>

* fix syncbn

* fix compare op in for in the cuda device

* fix the paddle compare op for the broadcast

* fix spawn default nprocs get error

* polish error message