* cherry picked

* "cherry picked platform"

* "add comment"

* "fix ci"

fix fluid_benchmark.py

* add flatten op

* refine regularization for selected_rows

* clean lookup_table

* refine rpc_server_test

* temporally disable rpc_server_test

* fix rpc_server_test

* add unit test

* refine conv cudnn enforce

* update

* update all cudnn ops

* fix

add doc

* Hiden some LoDTensor ralatted ops' Python wrapper.

* hide variable API

* edit API.spec