* Try to enable the runtime_context_cache pass in train phase.

* Put the append of runtime_context_cache pass ahead of multi_dev passes.
test=develop

* modify the interface of Pass::Allay
test=develop

* Polish code
test=develop

* Fix Travis CI
test=develop

* fix Pass::Apply interface
test=develop

* Fix Travis CI
test=develop

test=develop

* Support Sync Batch Norm.
* Note, do not enable it in one device.

Usage:

build_strategy = fluid.BuildStrategy()
build_strategy.sync_batch_norm = True
binary = fluid.compiler.CompiledProgram(tp).with_data_parallel(
        loss_name=loss_mean.name,
        build_strategy=build_strategy)

test=develop

test=develop

test=develop

* Choose to turn on use_mkldnn attribute v1

* Fix mkldnn_op empty bug

* format change test=develop

* fix ci test=develop

* fix ci test and add test in dam test=develop

* add example to dam compare test test=develop

* review changes test=develop

test=develop

test=develop

test=develop

Fix code to support cpplint syntax check

test=develop

* add MKL-DNN placement pass

This patch also refactors conv+bn (includes changes from PR
https://github.com/PaddlePaddle/Paddle/pull/13926)
updated to use the mkldnn-placement-pass.

test=develop

* remove redundant pass list

* add comment on the default first pass

* fix test for conv+relu mkldnn fuse

    Node should be created by Graph::CreateXXX so that
    they are managed by graph.

* dist transpiler add control dependency var between send and recv

* fix async deps

* follow comments and refine

* fix deps connect for rpc ops

* make enforce a target and dependent on nccl when gpu is enabled

* add some more dependency

ISSUE=4586495

git-svn-id: https://svn.baidu.com/idl/trunk/paddle@1408 1ad973e4-5ce8-4261-8a94-b56d1f490c56