Make code more clearer

* Clone method will create a new object instance, which is as same as
  itself.
* This is the first step to remove shared_ptr for OperatorBase

Using constructor to create an operator.

port gemm to new framework

Refactorize framework/*.proto

Merge branch 'refactorize_framework_proto' of https://github.com/wangkuiyi/Paddle into refactorize_framework_proto

Polish Our code by YuYang's review

enable self-defined index data in testLayerGrad.

Fix macro !defined(PADDLE_ONLY_CPU)

Rnn forward result test

Fix docker dev startup bug

Update Develop Branch And Fix All unit tests.

* remove SliceCopied

* remove SliceCopied

* rename SliceShared to SliceLevels, SliceInLevel

* merge lod_tensor/details

* remove lod_start_pos_'s shared_ptr

* make lod-tensor a special tensor

* add clone to lod_tensor

* add lod tensor impl

* add lodtensor clone test

* init lod

* add equal

* merge LOD and its methods

* recover tensor and variable

* change thrust to host_vector

auto update the requirements in .travis.yml with python/setup.py.in

fix cpplint error

add auto gradient check design doc

remove eigen tensor header file in ddim.h

Refactorize grad op builder.cc