Rewrite tensor.eye implementation and kernel

[ndgrigorian]

This PR proposes an update to the tensor.eye constructor, shifting to an approach using a 2D nd_range kernel which instead uses the condition that col - row == k along the diagonal, which is correct whether the array is C- or F-contiguous

This eliminates any branching in the kernel, in hopes of improving performance even marginally.

• Have you provided a meaningful PR description?
• Have you added a test, reproducer or referred to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• Have you checked performance impact of proposed changes?
• Have you added documentation for your changes, if necessary?
• Have you added your changes to the changelog?

[github-actions]

View rendered docs @ https://intelpython.github.io/dpnp/pull/2937/index.html

[coveralls]

coverage: 78.336%. remained the same — update-eye-ctor into master

[github-actions]

Array API standard conformance tests for dpnp=0.21.0dev0=py313h509198e_50 ran successfully.
Passed: 1357
Failed: 3
Skipped: 16