Refine profiler and expose to Python.

Fix LoD calculate of im2seq

Add pyton wrapper for row conv operator.

Conflicts:
	python/paddle/v2/fluid/layers/nn.py
	python/paddle/v2/fluid/tests/test_layers.py

Make Parallel.Do support sparse gradients 

Use ExternalProject_Add to download and install boost.

Fix warnings in priorbox, sequence_reshape and ctc_align_op

Enhance image.py for gray image.

* Add csp.md

* update csp.md

* Adding the distributed implementation for machine translation

* re-running CI

* Updated the code style

add output to fit a line demo

neural machine translation model support beam search with while op

Change default value of drop_rate in img_conv_group to 0

Enhance lookup_table_op to support padding_idx.

Fix the arguments transfer in image.py.

add iou similarity operator

Add python API for im2sequence op

* limit variable type to lod tensor in memory optimization transpiler

* refine policy

* support while operator

* fix random seed and training data order

* refine get_cfgs method to support multi while operators

* refine codes

Disable the unit test ofdetection output op