[v1.x] ONNX support adaptiveAveragePooling2D and update Softmax to support temperature

[waytrue17]

Description

Support onnx conversion of _contrib_AdaptiveAvgPooling2D. Update onnx conversion of softmax to support temperature attribute.

Due to some incompatible issue of mx-onnx conversion, currently we only support the conversion when output_size==1. Pytorch has the similar implementation: pytorch/pytorch#9711.

By inspecting all models in gluon model zoo, 16 of them has _contrib_AdaptiveAvgPooling2D. Out of the 16 models, 7 models use output_size>1, which are not supported in this pr.
Unsupported models:
psp_resnet50_ade
psp_resnet101_ade
psp_resnet101_voc
psp_resnet101_citys
psp_resnet101_coco
icnet_resnet50_mhpv1
icnet_resnet50_citys

Checklist

Essentials

• PR's title starts with a category (e.g. [BUGFIX], [MODEL], [TUTORIAL], [FEATURE], [DOC], etc)
• Changes are complete (i.e. I finished coding on this PR)
• All changes have test coverage
• Code is well-documented

Changes

• Feature1, tests, (and when applicable, API doc)
• Feature2, tests, (and when applicable, API doc)

Comments

• If this change is a backward incompatible change, why must this change be made.
• Interesting edge cases to note here

[mxnet-bot]

Hey @waytrue17 , Thanks for submitting the PR
All tests are already queued to run once. If tests fail, you can trigger one or more tests again with the following commands:

• To trigger all jobs: @mxnet-bot run ci [all]
• To trigger specific jobs: @mxnet-bot run ci [job1, job2]

---

CI supported jobs: [unix-cpu, centos-gpu, sanity, centos-cpu, windows-cpu, miscellaneous, website, clang, edge, windows-gpu, unix-gpu]

---

Note:
Only following 3 categories can trigger CI :PR Author, MXNet Committer, Jenkins Admin.
All CI tests must pass before the PR can be merged.

[waytrue17]

@mxnet-bot run ci [unix-cpu]

[mxnet-bot]

Jenkins CI successfully triggered : [unix-cpu]