Sequence Scatter Op

* Add python api for expand op.

* Fix unitest.

* Remove 'out' from arguments and fix code style.

* fix API.spec

* Fix API

* Fix unitest

* loosen the restriction of output_size in conv2d_transpose

* test and docs

* fix code style

* fix ci test error

* bug fix

* fix python3 issue

* refine seq_softmax_op

* fix seq_softmax

* use cub in seq_softmax

* Add sequence_expand_as_op

* follow comment

* add ignore index

* update api.spec

* enhance softmax_with_cross_entropy

* Add intermediate

* fix flatten/squeeze/unsqueeze

* Considering compatibility issues, we could not fix the origin op

* follow comment

* reset the shape of XShape

* Add pad2d op.

* Add unitest and python api.

* Fix cuda op kernel.

* Fix python api.

* Fix python api.

* Update API.spec.

* Fix python api

* enhance pad_constant_like_op

* add API

* add API

This reverts commit f5d5d7b2.

* Disable in_place in batch_norm API.

* Fix scatter_op python API and remove inconsistency between implementation and doc

* API spec change

* Change as review comment

* Add three modes for prelu_op.

Add flatten op interface and enhance APIs about detection to support variable-length image. (#12422)

* add flatten api&enhance detection api

* unify shape_op data type

* update API.spec