Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into softmax_cross_support_high_rank

Fix merge_lod_tensor_op infer shape

Reduce the layer number of transfromer model

Fix bug: long vgg-utest testing time 

Fix concat shape

Fix sum infershape issue

test=develop

fix split_byref_op infer shape

* - Reuse of conv PD

- conv transpose pd reused

- Added PD reusing of softmax and Batch Norm

- Refactoring and removal of not needed routines of mkl-dnn ops

test=develop

- Fix to reusing conv

test=develop

- Lint fixes

test=develop

- Further lint fixes

test=develop

- Lint  fixes

test=develop

- lint fixes

test=develop

- Lint workaround

test=develop

* - Fix after review on including boost as third party header

test=develop

* - Fix after review. Name change to something more descriptive

test=develop

* optimize lstmp and sample_logits op, test=develop

* update op_use_default_grad_op_maker.spec, test=develop

* delete useless file,test=develop

* append 0 to dim variable to avoid memory reusage, test=develop

* test=develop, fix hsigmoid dereference nullptr

* test=develop, refine condition

* test=develop, refine comments

infer shape compatable -1. test=develop

supprt high rank

Fix conv_shift_op infershape

test=develop

test=develop

Fix split

Fix auto growth bug of python dict in dygraph mode

add continuous value model op

if dim is -1, compile time check fails.

test=develop
Signed-off-by: Nzhaoyuchen <zhaoyuchen01@baidu.com>

fix memory optim temporarily

All any

* add SaveOptimModel interface in analysis_predictor.h and test it in analyzer_dam_tester and analyzer_resnet50_tester test=develop

fix the hang bugs of memory copying. test=develop

test=develop

test=develop, split complie option by internal and exposed

Reduce test_calibration test time by using large batchsize

Optimize merge add

test=develop

Add async sparse param update recorder

test=develop