* fix range in data flow analysis

Improve doc generation speed

Remove cpplint in cmake

Blas optimized elementwise_add forward and backward passes

Simplify label_semantic_roles book example using new LoDTensor API

Modify understand sentiment example using new LoDTensor API

Simplify Machine Translation demo by using Trainer API

Fix rst format issues in several files.

* Update label role implementation to use new API

* Try trainable embedding layer

* Fix feed order

* Add infer call

refine fluid deploy on teamcity

Fix ElementwiseOpInferVarType in elementwise_op

Disable and fix tests on multi devices.

Add `op_role` into OpDesc.