* Add Dockerfile for cuda9 and cuda10

Add Dockerfile for building cuda9 cuda10 images.

test=develop

update trt in  docker file
test=develop

test=develop

update trt in  docker file
test=develop

test=develop

* Refine Dockerfile

* Add tasks, cmake gen

* Fix code error

* Disable compile after paddle_build.sh

* Refine

* Skip on PY35 CI

* Change env

* Refine paddle_build.sh

* Expose gen_fluid_lib

* Refine mkldnn.cmake

* Refine mkldnn.cmake

* Refine mkldnnlib

* Skip unstable tests

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

Move six.moves.queue to Queue

* use -rpath to fix libmklml_intel.so not found

* only remain $ORIGIN/../libs in rpath of core.so

* test

* test

* add rpath of libmkldnn.so.0

* check return value of os.system

* remove check return value of patchelf

* 1. Remove PYTHON_FLAGS from paddle_build.sh in paddlepaddle/paddle:latest-dev

* 1. Add PYTHON_FLAGS back
2. Change SIGKILL to SIGINT and SIGTERM

* 1. Add setup.py.in back

* 1. add pip install open-cv in Dockerfile to avoid libusb_exit hanging up which is caused by the opencv-python package missing

* 1. Add the && \ to line above

* 1. Remove the notice comment

* scripts: clean bash scripts.

* Fix build related documents.

It seems that the `doxygen` package will remove system includes...

It should be a bug of ubuntu or docker. Since we are not using doxygen
now, just remove this package.

* use latest pip version

* fix ci

* update by comment

* "re-commit "

* "picked up"

* "fix ci"

* "fix pdb hang up issue in cuda 9"