* Fix recurrent op eager deletion pass error in dy2st

* Polish code

* Refine error message

* Rename BOOST_GET macros

* Fix conflicts

* boost::variant -> paddle::variant

* boost::variant.apply_visit -> paddle::visit

* Update pybind_boost_hraders.h

* Fix CINN compilation errors

* Revert FetchResultType

* Refine Constructor logic of ParallelExecutor

* Replace executor into ParallelExecutor in run_program_op

* use iwyu clean include, test=develop, test=win

* compilation error, test=develop

* fix compilation error2, test=develop

* fix compilation error3, test=develop

* fix compilation error4, test=develop

* fix compilation error5, test=develop

* fix compilation error6, test=develop

* fix compilation error7, test=develop

* fix compilation error8, test=develop

* fix compilation error8, test=develop

* fix compilation error10, test=develop

* fix compilation error11, test=develop

* polish reader error message, test=develop

* fix detail error, test=develop

* reset activation dcudnn change, test=develop

* add new macro BOOST_GET_SAFELY & unittests, test=develop

* add different macro type, test=develop

* fix get macro type in executor, test=develop

* four macro part change backup

* using one macro for all case, test=develop

* revert attribute change, test=develop

* change to three func to solve gcc4.8 bug, test=develop

* polish some details, test=develop

Test PaddingRNN on V100 GPU device.

Test configuration: large model, padding mode (which is the mode using recurrentOp), one GPU.
                   
GPU memory (MiB):   6414 (this PR)     vs   6837 (without this PR)
Speed (steps/s):         10.28 (this PR)    vs    9.89 (without this PR)