1. finish lookup table CPU and GPU kernel
2. Add some cuda helper
3. Add some math funtor

* Add Get/Has methods to OpInfoMap
* Add PADDLE_ENFORCE for OpInfo to get field.

Fix cycle dependencies, Fix #3583.

Fix #3582

* Also simplify pybind implementation by using OperatorBase as holder
  type.

Fit google C++ style

Make code more clearer

* Clone method will create a new object instance, which is as same as
  itself.
* This is the first step to remove shared_ptr for OperatorBase

* insert rnn's backward into Backward()

* add device_context into backward_test