* dynamic send/recv selected rows

* update by comment

* fix by comment

* Fix constructor bug in mixed_vector

* Fix warnings

* Clean code

* Extract for-loop init. Make nvcc happy

* dist train support lr decay

* update by comment

* revert elementwise method creator

* delete comment

* Refine target_assign_op to unify the classification and regression targets assignment.

* Fix the unit testing.

* Fix conflicts.

* fix absolute include path

* Remove test_helper.h in old location

* update include path

move paddle/pybind/pybind.h to paddle/fluid/pybind/pybind.h, and cancel the test_parallel_op temporary