Experiment on vgg flower, 2 trainers, 1ps.
    more trainer could have more speedup.

    After:
    Pass = 0, Iters = 327, Speed = (7.52) img/s
    Before:
    Pass = 0, Iters = 385, Speed = (6.77) img/s

add title for cpu_profiling.md

Reuse of pooling mkldnn primitives

Fix a compile error

* package prior_box op

* add doc

* add unittest

* add unittest

* fix CI fails

* Dockerfile: unify dev and product docker environment.

* Fix check style error.

* Use envrioment valuable to get version info.

Fix a profiler race condition

* use latest pip version

* fix ci

* update by comment

allow inference test to generate timeline

Inferencer support parallel_executor

fix roi_pool op bug

* listen_and_serv use localscope

* fix ut

Fix timeline for the number of profile is one

refine mkldnn cmake with official commit id

use cdn to speed up thirdparty package download

- Finished draft of pooling reusing of operators

- Using gethash in PoolGrad added

- Removed diagnostic

- Added pool mkldnn grad reusing of primitives

- Added diagnostic

- Removed diagnostic

- added dependency to mkldnn data type for pooling mkldnn

- Added mkldnn memory data type determining based on template type of op

- Compilation warning fix

- codying style fixes

* fix cmake error

* update fix

add mkldnn to paddle lib