* polish train eval faq

* polish details

* add paddle new save load doc

* move static save apis & polish multiple details

* add api mapping

* move apis place

* fix example code error

* polish model save load guide

* polish example code

* fix details example error

* polish guide doc

* change title format

* remove history reason

* remove other history reason

* add enable_static for old save/load

* add enable_static for old api

test=develop

* Add example for buffered and cache, test=develop, test=document_fix

* up, test=develop;test=document_fix

* 2.0.0-beta infer lib

* rename infer lib to paddle inference

* fix some docs

* fix links

* fix sample code

* test=document, bug fix

* test=develop, add utilbase doc

* test=develop, bug fix

* test=develop, ci

* test=develop, ci

* test=develop, rm fs

* remove paddle.reader APIs. test=develop

* remove paddle.reader

* optimize some words in image_segmentation/hight_level_api/getting_started

* fix image url error in image_segmentation

* link fix

* upgrade guide added

* update migration tool page

* minor change

* minor change

* update codes and format

* no ci for upgrade_guide

* minor

* minor

* add dygraph cn_doc;test=develop

* update doc

* add fleet meta lars and lamb api document

* add fleet meta lars and lamb api document

* test=develop

* update doc for hapi.model.save and hapi.model.Model in 2.0-RC

* fix example code for Model

* fix doc for Model example code

* fix sample code, add import InputSpec

* polish data read faq

* polish details

* remove static mode fag

* add fleet cn_doc;test=develop

We think we should split the usage and architecture because of clear organization and future extension (we may add more usage guide)

* fix windows pip install

* change =

* change =

* fix die link

* fix

* migrate en doc from old dir

* fix typo

* fix typo

* fix typo

Co-authored-by: Nroot <root@yq01-qianmo-com-255-129-13.yq01.baidu.com>