* PTQ save quantized model

* Wrap simulated layer

* post process the inference model

* opt dygraph python API, test=develop

* Fix unbind bug in manipulation.py

* reduce sum op default fp32, add into amp black list

* reduce_sum default fp32 can avoid return inf when the sum value large than 65504

* refine ptq according comments
* reuse the module to calculate kl threshold

* Save all scales to target ops
* Move quant layers to paddle.nn.quant

* refine the old code

* support moving_average_abs_max and per_channel_abs_max

* Add moving_average_abs_max_scale op

* Convert the test program

* dygraph post training quantization

* refine the ptq config

* refine ptq quantizer

* Combine amp and qat
* add unit test

* del py2 code2

* fix test timeout

* Add wrap for functional api
* Refine the wraped api
* Add unit test for quant functional layers
* Update all unit tests for dygraph qat

* support ssd in PsCore

* remove log

* remove bz2

* defalut value

* code style

* parse table class

* code style

* add define

* Update paths to Quant models

* Update description

* Add casting initializers for bf16 training

* Changes after review

* Correct test and add comment

* refactor_check_finite_and_scale_npu_kernel

* fix compile

* add alloc_float_status op

* add alloc_float_status op

* add FloatStatus for check_finite_and_unscale

* refine code

* remove unneccessary logic

* refine for fleet

* support ernie trt-int8 for inference

* fix reshape