[Cheery-Pick 2.0.0-rc1][Dy2stat] Add a decorator paddle.jit.not_to_static to support that not to convert a function in Dynamic-to-Static. (#29253) (#29340)

Usage scenarios：A function could have run successfully in static mode,  you can use it to decorate a function in the following cases:
  1. An unknown error occurs in the dynamic-to-static conversion process of the function;
  2. In the internal implementation of the function, it has two branches: dynamic branch and static branch;
  3. Users don't want to convert the function in the process of dynamic to static.

* use has_grad instead of train_mode

* add vlog for debug

* fix ut

* fix ut

* move temporal_shift to functional

* Change the api of DataParallel and Fleet (#29224)

* Add pure fp16 training with master weights. (#27712)

* add the weight decay func for the momentum op

* Add the multi_precision function in Momentum Optimizer.

* Make sure that the initial value of master weights are same with the fp16 weights.

* add static loss scaling.

* add the rescale_grad function in the pure fp16 training.

* use the original momentum updating method.

* Polish some codes, such as variable names.

* add docstring for apis.

* update the var creation details of _create_master_weight.

* not modify codes about imperative momentum updating.

* Fix the error of test_dist_sparse_tensor_load_momentum UT.

* add unit test for multi precision fp16 training.

* add more unit tests for CI.

* Use lower threshold values for allclose comparing in test_multi_precision_fp16_train UT.

Cherry-pick of PR #29226

* hot fix complle failed in gcc4.8

* fix failed unittest

* add lamb optimizer and unittest

* fix momentum resume training

* fix momentum acc

* add complex64 and complex128 type; add +-*/@ and slice opreator for complex types

* add test cases for complex elementwise, matmul and getitem unittest

* add test cases for complex types

* add test cases for complex matmul unittest

* The leaf tensor concept is exposed and the gradient accumulation of leaf tensor

* The leaf tensor concept is exposed and the gradient accumulation of leaf tensor

* fix coverage

* fix api doc

* fix CI unittest

* fix CI unittest

* fix unitest

* empty tensor does’t need inner_var_

* fix some error message

* fix paramete prefetch & device guard
Co-authored-by: NMrChengmo <cmchengmo@163.com>
Co-authored-by: Nchengmo <chengmo@baidu.com>

* Add a class TensorInplaceVersion to count the inplace version and put it in framework::Tensor instead of Allocation or Variable.

* Add a new attribute `_inplace_version` for VarBase.

* Raise exception if an inplace operation can result in incorrect gradient computation.

* Add a new interface _bump_inplace_version() for VarBase to bump the version whenever the Tensor is modified through an inplace operation.

* For api assign, call _bump_inplace_version() when it's an inplace operation inn dynamic mode.

* Use original var_wrapper if the inplace_version is not changed.

* Replace SnapshotVarWrapperList with SnapshotVarWrapper to optimize performane.

As the title

* fix failed tests in yingchun gived list

* add unittests into static_mode_white_list

* add enable static

* fix dist unittest

* skip test_sigmoid_focal_loss_op & add gym

* revert no need skip unittests

* remove gym

* Changed a variable name error

* Add comments

* Move member functions of TranslatedLayer out of function

* edit code according to review

* Edit input argument of '_run_static_graph'

* reset due to Segmentation fault

* rename variables when stitching graph

* modify code according CI

* Add comments to '__i_m_p_l__'

* remove blanks befor 'Get...'

* edit code according to review

* Add a comment to '_execution_method_creator'

* Edit a comment to '_execution_method_creator'

* Generate code coverage reports only for incremental files, test=develop

* Generate code coverage reports only for incremental files, test=develop

* Generate code coverage reports only for incremental files, test=develop

* test for diff python file, test=develop

* fix no python diff report, test=develop

* add cc test file, test=develop

* fix bug in generic.cmake, test=develop

* for debug no cc report, test=develp

* modify compire branch form test_pr to test, test=develop

* fix bug, test=develop

* test for h file changed, test=develop

* debug for redefinition of argument optimize error, test=develop

* close -o3 for test, test=develop

* remove -o3 for test, test=develop

* remove coverage option for nvcc, test=develop

* use CMAKE_CXX_FLAGS open coverage option when header file changed, test=develop

* reopen -o3, test=develop

* remove debug code, test=develop

* remove unused code, test=develop

test_mnist failed on CUDA11. We found that it is due to PaddleInference IR Optimization after debugging. We disable it in this PR and we will re-enable it after PaddleInference fixes it.

[Dy2Stat] Fix bug: the return statement should be transformed to an equivalent Paddle/Python if statement, which depends on if conditions of the return stmt. (#29165)

GridGenerator model failed because the output shape of `linspace` is (-1). The reason is that C++ InferShape fixes the shape to (-1): 

https://github.com/PaddlePaddle/Paddle/blob/5da3d514ebaa6fffd48c4a2e6bb5b16268dae92e/paddle/fluid/operators/linspace_op.cc#L49

We cannot set the shape in C++ infer shape because this Tensor may not be initialized during compile time, but when input `num` of `linspace` is an integer, we know the shape at compiler time. This PR simply set the shape in Python and add GridGenerator as unittest.
 

* add reducer

* refine envent for memorycopy

* add concat&split for allreduce

* apply concat & split for fuse tensor

* fix nccl dep

* fix the untest, compile problem and ddp initialize problem

* fix untest for mac & add some comments & solve the repeated param in sublayers

* fix untest for windows & fix document

update expand as op to use the shape of the target tensor instead of the target tensor itself. (#29020)

* update, test=develop

* alias yolo_loss & decode_yolo_box to paddle.vision. test=develop

add paddle.subtract, optimize paddle.maximum and paddle.minimum 

Add eigen gru and fix the dropout bug in the rnn 

[Dynamic-to-Static] Support **kwargs as input of the function which is decorated by `jit.save.to_static` (#29098)

* remove -DSUPPORTS_CUDA_FP16 in cuda.cmake

* comile with cuda9

* add some unittest

* notest;test=coverage

* add unittest for trt plugin swish && split

* update ernie unittest

* fix some error message

* remove repeated judgement of CUDA version in mbEltwiseLayerNormOpConverter

* fix comile errror when CUDA_ARCH_NAME < Pascal"

* fix comile error

* update unittest timeout

* compile with cuda9

* update error msg

* fix code style

* add some comments

* add define IF_CUDA_ARCH_SUPPORT_FP16

* rename IF_CUDA_ARCH_SUPPORT_FP16 to CUDA_ARCH_FP16_SUPPORTED

* add symlink force for unittest

* open unittest

* update, test=develop