* upgrade fp32 to fp64, remove no_grad_set, enlarge input shape for hsigmoid unittest test=develop

* revert "remove no_grad_set" test=develop

* fix input shape of op tests for elementwise_sub,
gather, pad2d, transpose, softmax, scale,
elementwise_max, hierarchical_sigmoid, reshape2,
sign, squeeze, reduce_sum, sum, squeeze2,
unsqueeze, unsqueeze2, cast, reverse

test=develop

* fix cast, elementwise_mul, gather, scale, sign,
softmax, transpose

test=develop

* "cherry picked cpp tests"

* "cherry picked"

* "cherry picked tests"

* "merge develop branch"

Fix the HierarchicalSigmoidGradOpKernel and refine the codes. Now hsigmoid_op is same with V2 implementation and can pass gradient check.