* run dy2stat pure fp16 in Linear model

* no use self._pure_fp16_inputs

* add test and fix Adam error in dy2stat pure fp16 training

* use paddle.optimizer.Adam

* run test in gpu

* change test time for CI

* enlarge atol for test_resnet_pure_fp16

* refine code and enlarge atol

* make custom_white_list and custom_black_list take effect for AMP and pure fp16

* check tracer is not None

* use default atol

* change filter_size

* change atol and add some NOTE

* Add EagerTensor and tests

* remove useless enforce

* remove comment in cmake

* support autograd meta

* support grad node info test

* support grad_node_info

* add more edge test

* remove Python.h

* add tensor wrapper with tests

* support compute require grad and stop gradient

* support sync methods and global utils

* support pure cpu test

* refine error msg

* refine error msg

* refine error info

* fix npu error

* fix inplace bug

* fix custom grad input error

* add unittest

* fix inplace bug

* [XPU] Reorganize xpu device codes in platform, test=develop

* fix xpu_header.h, test=develop

Add support for bias is none for fused_attention op.

* modify code about fp16 of dcnv2 trt

`paddle.utils.download` ：change to call `extractall` on tar/zip compressd file  to speed up the uncompress process when they includes many files

--- result of decompression speed comparison ---
1. dataset：https://paddlenlp.bj.bcebos.com/datasets/cnn_dailymail/cnn_stories.tgz, decompression time
：5m50s vs 20s
2. dataset：https://paddlenlp.bj.bcebos.com/datasets/cnn_dailymail/dailymail_stories.tgz, decompression time：33m20s vs 47s

* skip compiled program with places > 1

* fix corner case and add ut

* sync scope and variable_scope when init executor

* set var_desc for new var

* test on_infer off problem in windows

* turn off on infer in windows-ci

Fix PR-CI-Static-Check

* fix ut retry

* fix_nearest

* fix_nearest

* fix_nearest

* fix_nearest

* for pure fp16

* opt topk

* modify for wincheck-inference case

* modify according to zhouwei's comment

* modify according to zhouwei's comment 2nd time
Co-authored-by: Nzhangkaihuo <zhangkaihuo@baidu.com>

* enhance scatter err msg check

* fix ci error

* elementwise_div refactor

* fix compile bugs in windows ci

* Add EagerTensor and tests

* remove useless enforce

* remove comment in cmake

* support autograd meta

* support grad node info test

* support grad_node_info

* add more edge test

* remove Python.h

* refine error code

* add error type in error msg

* given default null name for tensor

* Added HCCL backend support in dynamic graph mode

* fix segmentation fault

* add ut

* Bug fix for snapshoting VariableWrapper with initialized tensor but empty allocation

* Added unittest for inplace&clear_gradient

* rename to api to copy_to

* support cast method for tensor

* fix compile failed

* adapt to inference api dir for pten

* fix conflit with develop

* fix test_egr_ds_eager_tensor compile failed

* Add transfer_layout/dtype op

* clean useless codes

* fix unused var

* add optest in white.txt

* split into data_transfer.cc

* fix cmake

* modify according reviewer comment

* replace cast_op with transfer_dtype_op

* disable copying of datatype when sharing buffer between two tensors.
* fix for mkldnn operator kernels (elementwise_add, sum, softplus, softmax, scale, activation), mannually set the data type when reusing memory by ShareBufferWith.

* fix autoconvert

* fix merge parameter

* add isclose op, test=develop

* add isclose op, test=develop

* add isclose api, test=develop

* rm useless code

* rm useless code

* update python api of isclose

* add some unittest of isclose op, test=develop

Fix CI build error

[Dy2stat]Allow users to switch eval/train mode when using @to_static to decorate a function (#37383)

* Allow users to switch eval/train mode when using @to_static to decorate a function

* refine code for train() and eval()

* support zero dim for slice op

* support zero dim Tensor in set_value op

* polish some debug log