* Polish executor design doc

* Adding few details

* Simplize system_allocator and fix GPU_INFO

* Remove main_program/startup_program

* Fix optimizer

* Fix CI

* Follow comments

Add python wrapper for SeqExpandOp

update mkl packed design doc

refine mkl&openblas script for benchmark

refine cluster doc structure

Rename seq to sequence in sequence_expand_op

Fix crop op doc

Fix error in function/GemmConvOp.cpp.

add spp（Spatial pyramid pooling ） op 

fix simple_gru2 doc

* "add DeviceContextPool"

* "add devicecontextpool in pybind"

* "add comments in python side "

* "fix static link error"

* "fix CI error"

* "add executor.py"

* "fix CI error"

* "add with gpu macro"

* "remove comment out codes"

* "add TODO items"

* "update init devices"

* add more place_test and rename Cudnn to CUDNN

* fix ci

Optimizer is not a part of fluid. The `paddle::platform`
was missing.

use Intel OpenMP to speedup seq2batch when WITH_MKL