Added implementation to calculate Gram and Row Gram matrices

[TwentyPast4]

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

Gram and row Gram matrix computations (ie. A.T @ A and A @ A.T) are relatively common in some linear algebra (eg. least squares, linear independence, ML kernels, ...).
If you execute A.T().Matmul(A), at least one of the two matrices in the matmul operation will not be contiguous, meaning when matmul is executed, a copy operation will be performed which can be a noticeable performance loss.
Implementations of Gram() and RowGram() are done with a single matrix, with the transposition being done in gemm functions. This means that if A is contiguous, no copy will be performed, which is not true for A.T().Matmul(A).

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

Added Gram() and RowGram() functions to Tensor. These are intended for <=2D tensors, similar to how T() is implemented.

[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.