Optionally include CUDA DLLs in open3d-package

[fegorsch]

Type

• Bug fix (non-breaking change which fixes an issue): Fixes #
• New feature (non-breaking change which adds functionality). Resolves #
• Breaking change (fix or feature that would cause existing functionality to not work as expected) Resolves #

Motivation and Context

In order to have a self-contained Python package with GPU-support on Windows, allow bundling the CUDA DLLs with open3d. This is especially useful when deploying open3d to non-developer machines where you cannot assume that CUDA is installed. It also is a step towards an official Python package with CUDA-support for Windows, which would save the effort of building yourself.

Checklist:

• [ x] I have run python util/check_style.py --apply to apply Open3D code style
  to my code.
• [ x] This PR changes Open3D behavior or adds new functionality.
  • Both C++ (Doxygen) and Python (Sphinx / Google style) documentation is
    updated accordingly.
  • I have added or updated C++ and / or Python unit tests OR included test
    results (e.g. screenshots or numbers) here.
• [ x] I will follow up and update the code if CI fails.
• [ x] For fork PRs, I have selected Allow edits from maintainers.

Description

Added CMake variable BUNDLE_CUDA_SHARED_LIBS, which causes all DLLs from the build system's CUDA-installation to be included in the wheel. Only works on Windows and only if BUILD_CUDA_MODULE is on.

[fegorsch]

Found that CUDA can also be installed via wheels available on pypi. Perhaps defining those as requirements is a better way of solving this, e.g. pytorch, Jax and tensorflow seem to resolve their CUDA-dependency this way as described here. I'll look into that and open a new PR.