Currently use `Operator("fc", X="x", W='w1', B='b1')` as operator
creation method.

Fix #3198

* init softmax grad op

* add compute code

* export Backward to python

* update test ,export op.type to python

* update python test, fix compute bug

* update unit test

* use eigen

* optimize eigen code

* add gpu test

* register softmax_grad GPU kernel and fix test bug

* typo

* follow comments