* get output when training

* follow comments

* using etcd as fault tolerant training

* update

* workable version, ft not tested

* small fix

* update

* remove TODO

* PyBind and SWIG of paddle cannot be load in a single Python process,
  lazy import all SWIG library of Paddle. Otherwise, the glog, gflags
  are imported twice in a same Python process.
* Note that all PyBind11 return C++ std::string as an unicode. For
  protobuf, it is need be cast to `str` before use them.
* Add unit test for Get `OpProtos`

* Also fix some other bugs.
* Fix #1495

remove ITrainer interface since parameters for train is already getting out of sync. Currently there is no benefit of having a interface.