* Changing learning rate from attribute to input(float)
* Removing obsolete code

The private data `tensor_shared_` is not used.

* add tensor array

* update

* set type

* fix cond_op_test and add some check to operator.run

* tmp

* optimize kernel check

Add multiple Inputs/Outputs check interface

* init remove cond_op infershape

* optimize code

* add PrepareDataForSubnet and MergeDataFromSubnet