* new usage of amp training.

* change the usage of amp and pure fp16 training.

* modified code as reviews

* support fp16 training

* Use compiled training program

* Change timing ips.

* Use dali

* add pure fp16 training

* fix a bug, which will not use fuse pass using pure fp16 training.

* modify code as review

* modify loss, so that it will use different loss when using pure fp16 training.

* remove some fluid API

* add static optimizer.

polish codes and docs to adapt Paddle2.0-rc