* add cudnn related strategies to DistributedStrategy

* optimized transformation form tensor to numpy, test=develop

* optimized transformation form tensor to numpy, pass pre-commit, test=develop

* modify fetchophandle zerocopy to deepcopy in PE&CUP, test=develop

* modify py:array construct, test=develop

* fix _fetch_var to use deep copy, test=develop

add the isnan, isfinite, isinf api for the paddle 2.0

* support Baidu AI Accelerator
  * test=kunlun

* minor
 * test=kunlun

* support xpu op in separate file
 * test=kunlun

* update XPU error message and remove duplicated code

 * test=kunlun

* minor
 * test=kunlun

* minor
 * test=kunlun

* polish framework error msg part 5

* revert enforce change

* refine error type

* trigger ci check

* polish details by review comment

* Conv2d_transpose and mul support channnelwise quantization, test=develop
* Skip collecting out threshold for output tensor of which the type is not fp32 or fp64, test=develop
* Fix error in test_user_defined_quantization, test=develop
* Add depthwise_conv_bn_fuse, test=develop
* Add conv_transpose_bn_fuse_pass for post_training_quant, test=develop

test=develop

* expose and unify the Tensor concepts to the user

* expose tensor to user

* add copy place for Tensor

* add copy place for Tensor

* add note

* add macro PADDLE_WITH_CUDA

* remove RUN_TYPE=DIST

* fix some error

* move paddle.fleet to paddle.distributed.fleet

* add unchaged infershape function

* add broadcast infershape function

* fix bug

* rename infershape functions

* add UnaryOpUnchangedInferShapeCheckAxis

* add error message

* add test for common infer shape functions

* dont update existed ops

* dont update op_desc.h

* add more test

* add error check, refine error message

* fix loaded var suffix repeat error

* use new dygraph name for loaded param

add lamb to fleet meta optimizer

* add encrypt for inference lib

* fix code;test=develop

* fix test; test=develop

* bug fix; test=develop

* add MakeCipher;test=develop

* fix bug;test=develop

* move MakeCipher to paddle space; test=develop

* fix include dir ;test=develop

* add include dir; test=develop

* move include; test=develop

* move include; test=develop

* fix for windows ci

* fix cmake; test=develop

* fix bug

bug fix

* add paddle.fleet.AsyncOptimizer
Co-authored-by: Ndongdaxiang <dongdaxiang@baidu.com>

* Change use_quantizer attribute name and data type

* Fix problem with setting attribute

* Add changes due to review

* Small change in function

* Restore use_quantizer attr for compatibility

* fix large scale KV 
* fix single training using async ssa graph

test=develop

* remove global variable

* refine code

* test=develop,test=document_fix, remove the out args

* fleet_util move to paddle.fleet
Co-authored-by: NWuHaobo <wuhaobo1994@gmail.com>
Co-authored-by: Ntangwei12 <tangwei12@baidu.com>

* add heter ps mode

* code style
test=develop

* add with_pslib
test=develop

* unitest
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* test monitor
test=develop

* prepare trainer
test=develop

* code style
test=develop

Add dgc to fleet meta optimizer, rm dgc from optimizer all

* fix dump, fix cvm check
test=develop

* fix
test=develop

* fix
test=develop

* fix
test=develop

* split meta optimizer files
* add graph execution in execution, update two properties in DistributedStrategy, unit tests for these features

* polish framework error message part 4

* fix type error

* fix message error

* polish by review comments

* fix test_pipeline, test=develop

* refine error stack format

* polish compile traceback format

* polish detail format

Integrated Trainer of Parameter Server (API add `fluid.contrib.layers.sparse_embedding` only) (#22957)

* Integrated Trainer of Parameter Server

* refine strategy compiler and meta optimizers
make async as a_sync

* simplify buffered reader to improve DataLoader performance

* fix 22 failed unittests

* fix cuda pinned context condition

* fix test_reader_reset failed

* fix two failed unittests

* change unittest place

* polish error messaage

* polish cast op GetExpecctedKernelType

* remove debug info in unittest

* add more settings for distributed strategy
Basically, DistributedStrategy has several parts of configurations:
- BuildStrategy: the same as paddle.fluid.BuildStrategy, but the distributed arguments are moved out of BuildStrategy
- ExecutionStrategy: the same as paddle.fluid.ExecutionStrategy
- collective communication configs: nccl_comm_num, hierarchical allreduce and so on
- distributed algorithms: async_update(mainly used in PS), lars, lamb and so on