* test=develop, bug fix

As the title, decrease random failure probability for test_parallel_executor_mnist

The old code set larger delta when comparing reduce and all reduce, but didn't set all. I added it.

On my linux machine, I run 100 times, no failure occurs. In addition, we only saw this random failure on CI two times since I worked. I thought it was rare and I just increased the delta.

* Remove dependency on nltk for paddle __init__.
test=develop

* Remove nltk.movie_reivew sentiment dataset to remove dependency on nltk.
test=develop

* reproduce summary api

* rename StaticLayer

* rename

* fix dtype of gradientclipbyglobalnorm

* fix dtype bug of GradientClipbyGlobalnorm

* init commit of new save/load

* fix failed unittests

* fix save_load_v2 unittest failed

* fix failed unittest & polish doc

* add tests for coverage

* add more tests & move static apis

* fix example code error

* polish emample code

* fix detail example code problem

* Makes the Bind Method more intelligent

* Makes the Bind Method more intelligent

* fix unittest

* fix unittest

* fix conflict

The calculation result of Diag_v2 Op under large size input is wrong 

* rename communicator meet->BatchesCounter

* fix parame recv for sparse

* geo sparse init from pserver

* optimize init from pserver

* add large scale optimizer fuse(SGD/ADAM)

* rectification init_worker and exe.run startup program

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

* add fused_bn_add_relu op

* default open dygraph mode

* fix CI-Mac

* fix Mac-CI other unittest file

* fix CI-Py3

* fix test_communicator_geo and test_buffer_shared_memory_reuse_pass

* add enable_static to fix CI-Py3

* add enable_static to fix CI-coverage

* delete try except

* refine error mesg

* remove paddle.readers API. test=develop

Add the cpu version of segment sum mean max min op

test=develop

* argsort op acceleration on GPU when the input size is equal to the length of the ‘axis’ dimension

* add double grad compute for batch norm,test=develop

* fix unittest, test=develop

* remove unuse tensor,test=develop

* add format,test=develop

* update, test=develop

* fix bug sequececonv_eltadd_relu_fuse_pass, output error when sequence_conv's padding_start > 0

* fix seqconv_eltadd_relu_fuse_pass unitest error

* support use add instead of sum to do gradient accumulation

* add inplace addto pass

* add grad_add op and inplace addto pass

* remove debug code

* code refine

* fix bug when sereral sum ops inserts at same op_idx

* fix Flags type

* add addto attribute for conv3d

* fix ut

* code clean

* fix type

Fix bug: shapes of Topk outputs are wrong when the parameter k is Tensor 

* Finished ChannelWiseQuantDequantAbsMaxOp and Passed unittests.

* Finished channel-wise quantize strategy in imperative quantization.

* Added Cuda code of ChannelWiseQuantDequantMaxAbsOP
Add Cuda code of ChannelWiseQuantDequantMaxAbsOp

* Add quant_axis for channel_wise quant.

* fixed a bug in unnitests, which will not trigger axis = 1 case and cannot meet the coverage rate requirement.

* Added some assert infomation and fixed some coding style mistakes.