Centerpoint bbox coders: Fix torch warning

[chbehrens]

Motivation

I currently get the following warning from the centerpoints bbox coder:

mmdet3d/models/task_modules/coders/centerpoint_bbox_coders.py:207: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
  self.post_center_range = torch.tensor(

The reason is that self.post_center_range is already a torch.Tensor after the first call of decode and then torch.Tensor is called on a Tensor in every following call.

Modification

Create self.post_center_range as Tensor in the initialization of CenterPointBBoxCoder and only make sure that it is on the right device when decode is called.

BC-breaking (Optional)

Does the modification introduce changes that break the back-compatibility of the downstream repos?
If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.
No breaking/functionally relevant changes

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here, and update the documentation.

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 test to ensure the correctness.
3. If the modification has potential influence on downstream projects, this PR should be tested with downstream projects.
4. The documentation has been modified accordingly, like docstring or example tutorials.

[Xiangxu-0103]

Should notice that the default post_center_range is None, which may cause an error here.

[chbehrens]

True, I updated it so that it stays None in that case.

[chbehrens]

I added another small commit to fix the type annotation for vel in the decode function as it can be None as well as the dimensions of the dim and vel tensors in the docstring.

[chbehrens]

Could someone please retrigger the pipelines that got stuck so this can get merged?