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

    paddle.save paddle.static.save protocol的默认值改为4（原默认值为2）。

    pickle protocol=4相交于protocol=2：
        protocol=4时保存/加载大于4G的单个numpy.ndarray 等。
        protocol=4时保存/加载的速度有明显提升。
        Python2 不支持protocol=4（paddle2.1主要支持Python3，不再考虑Python2）。

    兼容问题：pickle版本（protocol）会写到文件里面，pickle load的时候会自动识别到protocol，paddle2.1（paddle.save pickle默认版本为2）可以加载paddle2.1.1的模型（paddle.save pickle默认版本为4）。

原始PR：#32904

Cherry-pick (#33213)

 [cherry-pick][CustomOP]Set GLIBCXX_USE_CXX11_ABI=1 to fix potential GCC ABI problem  (#33153) (#33185)

* Add GLIBCXX_USE_CXX11_ABI flag

* fix typo

* fix typo

* [HybridParallel]Fix pipeline in dygraph (#33007)

* fix pipeline

* fix mp pp dp

* fix utest of hybrid parallel

* add utest for tuple

* fix utest (#33108)

* Support convert sublayers in Sequential Container

* remove paddle.jit.set_code_level

* fix precision of mp

* fix bug of seed

* fix dp

* print group