Fix a bug in NonZeroCUDA not using the correct memory addresses

[TwentyPast4]

When using any advanced indexing operations like IndexAdd, IndexSet, IndexGet with a boolean mask, the results are incorrect in the following case:

• the tensor is on a CUDA device, and
• the _data_ptr of the tensor is not equal to the start of the Blob data pointer (e.g. as a result of slicing)

The problems seems to be that inside NonZeroCuda.cu, GetBlob()->GetDataPtr() is being used, instead of just GetDataPtr.
There seems to be no other place such usage occurs, and other CUDA functions use Tensor::GetDataPtr(), so it seems to me it's very likely this is a bug.

See issue #7341

Type

• Bug fix (non-breaking change which fixes an issue): Fixes Advanced indexing with boolean masks on CUDA can be incorrect #7341
• 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 my application, I was using something like this (pseudocode):

Tensor a = someDataSet; // let's say this is of shape {N} or {N, 1}
Tensor mask = a.Ge(0);
Tensor subsetA = a.Slice(0, 10, 20);
Tensor maskSubRange = mask.Slice(0, 10, 20);

Tensor maskedSubsetPart = subsetA.IndexGet({maskSubRange}); // here NonZeroNumpy will be called to convert the boolean mask to a list of indices, but it will use the blob's data pointer instead of the tensor's data pointer as the start of the search
Tensor maskedFullSet = a.IndexGet({mask});

// now I would expect this to be equal, but it's not
EXPECT_EQ(maskedFullSet.Slice(0, 10, 20), maskedSubsetPart)

Checklist:

• I have run python util/check_style.py --apply to apply Open3D code style
  to my code.
• 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.
• I will follow up and update the code if CI fails.
• For fork PRs, I have selected Allow edits from maintainers.

Description

[update-docs]

Thanks for submitting this pull request! The maintainers of this repository would appreciate if you could update the CHANGELOG.md based on your changes.