* add pure fp16 major function in auto_cast & tracer

* support master weight in dygraph for pure fp16

* check mix dtype of fp16&fp32 for check_finite_and_unscale op

* change pure fp16 funtion name

* refine some bug in auto_cast

* refine auto_cast interface logic

* add param _casted_by_pure_fp16 for class Layer

* support state_dict hook for save model by user appointed dtype in pure_fp16_decorator

* refine pure_fp16_decorator as decorator

* add unittest

* add comment

* add comment

* support recompute

* add comment for auto_cast and decorator

* support to_static_state_dict for paddle.jit.save

* unlimite models num and optimizers num

* add lookup_table in black_list

* fix momentum and layer state_dict

* fix bug in layer state_dict

* fix bug in layer state_dict_helper

* refine unittest

* refine test_momentun_op

* refine interface and some code

* refine amp_decorator interface

* refine pure fp16 interface

* refine master weight interface

* temporally disable the warnings

* disable ut

* add pinv api, test=develop
* add linalg pinv api, test=develop
* update example code, test=develop

* Support EMA in Paddle2.x and Fleet

* update

* update

* update

* modify ut of ema

* modify docs

* modify bugs

* update

* update

* update

* modify ut

* add scale_op in model save step is not necessary, just fix the prune method to support static graph and inplace op

* fix jit.save, no need to add scale_op to each outputvar anymore.
fix prune_with_input, now it supports inplace op

* temporarily disable test_trt_dynamic_shape.TRTDynamicShapeOutOfBound2Test

* add_skip_layernorm

* add_skip_layernorm

* add_skip_layernorm

* add_skip_layernorm

* add_skip_layernorm

* add_skip_layernorm

* add_skiplayernorm_teller

* add_skip_layernorm

* add_skip_layernorm_teller

* add_skip_layernorm_teller

* add_skip_layernorm

* add_skip_teller

* expose cuda stream to users

* add ut

* add test

* add test

* add test

* add test

* add test

* add test

* add test

* add test

* fix unpool doc, test=document_fix

* fix typo for python example, test=document_fix

* update equal op, input Y can be float,int,bool or tensor

* update test

* update code style

* update code style

* update doc

* update str check

* remote str

* add type check

* add launch doc

* add trainer desc config to distributed strategy

* code style modified

* data_feed set lod

* fix bug

* code style

* fix bug

* [NPU] add index_select_grad kernel and unit tests

* dim=0 not need transpose

* fix DataLoader exit with SIGABRT/SIGSEGV. test=develop

* Support new API linalg.cond in paddle

* check code style

* check code style

* modify codes

* add docs_eng of linalg.cond

* add svd_norm for linalg.cond

* modify docs_en of cond

* add support for empty input in dynamic mode

* modify set_time of unittest

* update

* modify unittest of cond

* update

* remove cond in paddle.__all__

* pull latest codes

* merge latest codes

* update

* remove autograd/grad api

* import grad, no_grad set_grad_enable from autograd module

* modify import no_grad_ as no_grad

* fix gather

* fix

* fix no_grad context error in dy2stat

* remove useless comments

* fix error by drop_kids in python

* add test and fix review

* support fp16 dtype

* update

PR主要功能：针对fusion等子图替换场景，支持Python侧开发并注册Pass。

背景
Pass是指输入一个深度学习计算图Graph，依照一定条件进行修改，输出修改后的Graph的过程；
当前PaddlePadle框架编写Pass代码存在以下问题：
用户需要手写Graph的条件匹配、在Graph上的修改代码；
对Graph操作需要深入底层框架代码，了解Graph的结构，并且知道相关Pass写法；
我们提出了针对fusion等子图替换类Pass的优化方案以支持用户在Python侧开发注册Pass，提升二次开发体验：
用户只需要输入匹配和替换的子图描述，由深度学习框架编写的代码来生成匹配和替换的逻辑，不需要用户对Graph进行匹配和替换操作；
API级别的替换，用户可以通过Paddle的Python API构造子图，从而不需要知道Graph的结构，也能写Paddle的Graph Pass代码

Fix docs in stream_guard API 

* fixed slice error

* added handling of StartsTensor+List and EndsTensor+List

* fix for ppyolo model

* add inplace logic into new_executor, test=develop

* check shape and add inplace FLAGS, test=develop

* refine, test=develop

* refine, test=develop

* clip op extra information when export model,test=ocr

* rename clip_extra parameter to kwargs in save_inference_model, test=ocr

* Change the invoking method of settiem from numpy to set_value op when value is not tensor

* fix the check logic for inplace in setitem

* fix the unittest problem caused by setitem doesn't support fp16

* modify some code format in setitem