[Feature] Add MinkUNet segmentor

[sunjiahao1999]

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

Motivation

Please describe the motivation of this PR and the goal you want to achieve through this PR.

Modification

1. Add minkunet voxelization
2. Add minkunet backbone
3. Add minkunet decode head
4. Add torchsparse optional backend

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.

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.

[filaPro]

Btw, do we want to set different names to resnet and unet models based on torchsparse (in this PR), minkowskiengine (in MinkResNet class and I'm also thinking on adding TD3D segmentor with MinkUNet) and spconv (in SparseUNet class)?

[ZCMax]

minor comments to be fixed and then I'll approve the PR