[Feature] Support centernet dev1.x

[hanrui1sensetime]

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily receiving feedbacks. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Support CenterNet which is P0 model.
But trt 8.4.x has bugs in dynamic shape configs. Refer to: NVIDIA/TensorRT#2299

Modification

• Rewriter
• UT and reg tests
• docs

BC-breaking (Optional)

None.

Use cases (Optional)

None.

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness.
3. If the modification has a dependency on downstream projects of a newer version, this PR should be tested with all supported versions of downstream projects.
4. The documentation has been modified accordingly, like docstring or example tutorials.

[RunningLeon]

@hanrui1sensetime pls. fix lint error.

[RunningLeon]

@hanrui1sensetime may fix conflicts.

[grimoire]

The result is different between tensorrt and pytorch

[grimoire]

And we better find a way to get rid of adding new config. @RunningLeon

[hanrui1sensetime]

The result is different between tensorrt and pytorch
May be caused by TensorRT issue 2299 NVIDIA/TensorRT#2299
Do you test different between tensorrt and pytorch on TRT-8.2.x too?

[RunningLeon]

@lzhangzz sdk ci failed, is there something wrong with changes of Pad?

[RunningLeon]

@hanrui1sensetime pls fix conflicts of docs

[RunningLeon]

Pls. correct the docs.

[RunningLeon]

LGTM

[RunningLeon]

@grimoire @lvhan028 Hi, guys, any other comment on this PR?