Fix CI error on test_LayerGrad.LSTM

* We should not EXPECT_EQ between a float value and a int value.
  Use ASSERT_NEAR instead.

Merge branch 'add_golang_precommit' of https://github.com/typhoonzero/Paddle into add_golang_precommit

fix bugs for CrossChannelNormLayer

fix cmake errors

Correct GLOG CHECK in Paddle

stride pooling for max and average layer

Use CHECK instead of PCHECK, because PCHECK is used for errno.

Remove buggy BarrierStat

fix compile errors in tensor. 
`PlaceholderImpl::Deleter` maybe used in other modules. We can consider moving it to be `paddle::memory::Deleter`.

fix cmake error

Add DetectionOutputLayer and MultiBoxLossLayer.

The implementation of BarrierStat is buggy, and it is not necessary
for Paddle to diagnose which node in cluster is slow.

* add op_desc.proto

In Operator design, we need a proto message to describe an Operator.
Third-party language such as python can build this proto message and use
AddOp(const OpDesc& op_desc) of Paddle core to construct an Op in the
Network.

fix wrong including header-file in files in paddle/platform/dynload dir