* code clean - refine ARM

cmake enhancement:

- add lite_cc_library and lite_cc_test

code clean:

- remove ARM feed and fetch kernels, reuse the Host's

remove unnecessary comments

* add cmake

* update

* fix proto pd

* fix compile

* tmp save

* fix protobuf device version

* fix protobuf and host compile

* fix std c++11 support on android

* change array to vector to fix ndk c++_static

* fix rt and add dockerfile

* fix android compile issue with latest merge

* init arm kernels

* enable run on arm

* update format

* update format

* update format

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

make typesystem simpler

* reset drop local scope counter
test=develop

refine code

fuse cublas  calling and kernels into one cuda kernel.

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

* Add conv2d_grad_grad_op
* Extracte the cuDNN conv algo searching code in conv_cudnn_helper.h.
    - Now use it in conv2d_grad_grad.
    - Will simply the searching code in conv2d and conv2d_grad in next PR.
* Enhance and fix bug in unit testing of gradient_checker.
* Support to fetch empty variables，return None in Python.

* rename the default version from '0.0.0' to 'latest'

test=develop

1. infernce multi card occupy
2. facebox model inference occupy too much
test=develop

* test=develop
add elementwise_mod and elementwise_floordiv, fix equation problem in elementwise_mod

* fix some logic in distributed transpiler, test=develop
* reformat fleet API, test=develop

The parameter dirpath will be passed directly to c++ operater. The file address format will be different under win and UNIX.

* fix train_from_dataset and infer_from_dataset example

* add inductive dim for data_reader, example: shape=[-1, 1], then -1 will be inducted through run-time reading of number of elements

test=develop

* Fix API example code of save_inference_model

test=develop

* Add "import" in exmaple of save_inference_model

* Fix typo "exsample" -> "example"

test=develop

* modified formula for lrn

test=develop

* modified api.spec

test=develop

* Refine elementwise kernel.

Add a simple cuda kernel if grad x and y both exist
Use 2D block cuda kernel to do broadcast.

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

* refine code.

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

* refine code.

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

Fix the following API examples:

paddle.fluid.scope_guard
paddle.fluid.backward.append_backward
paddle.fluid.cpu_places
paddle.fluid.cuda_pinned_places
paddle.fluid.cuda_places
paddle.fluid.in_dygraph_mode
paddle.fluid.CUDAPlace
paddle.fluid.CPUPlace
paddle.fluid.CUDAPinnedPlace

* Optimize the cuda implementation of sum_op, which add two lod_tensors inplace.
test=develop

* Use eigen to add to tensors.
test=develop