feat(label_tensor): add dim parameter to LabelTensor.stack method

[s-chudmunge]

Resolves #770

Summary

Exposes the dim parameter in LabelTensor.stack(tensors, dim=0) to align with torch.stack.

Key Changes

• Added dim=0 argument to LabelTensor.stack with support for positive and negative dimension indices.
• Validates bounds [-(ndim + 1), ndim] matching torch.stack.
• Dynamically updates _labels dimension indices by shifting indices at or after dim by +1.
• Validates consistency of input tensors and their stored labels.
• Added comprehensive unit tests in tests/test_label_tensor/test_label_tensor.py covering dim=0, dim=1, and negative indexing.

[dario-coscia]

Hi @s-chudmunge ! Thanks for submitting. Are you able to rebase from dev? I see some commits that are not labeltensor based from this PR after changing the destination merge branch

[dario-coscia]

Hi, great job! I added a comment on the test for labeltensors to also check additional dimensions where the stack is performed. After that addition I think it will be good to go 👍🏻

[dario-coscia]

Can we make an extra test here to check all labels' dimensions and not just the last one?

[GiovanniCanali]

Thank you, @s-chudmunge, for contributing to PINA. The feature looks clean!

I agree with @dario-coscia’s suggestion: please rebase the branch onto dev so that the PR includes only the commits relevant to LabelTensor.

Also, consider adding tests covering higher-dimensional cases, similar to test_vstack_2D and test_vstack_3D.

[s-chudmunge]

Hi @dario-coscia @GiovanniCanali,

Thanks for the feedback!

1. I have rebased the branch onto dev so that only the commits relevant to LabelTensor are included.
2. I have expanded the unit tests with test_stack_2D and test_stack_3D covering higher-dimensional tensors across all positive/negative dimension indices and verifying full_labels on all dimensions, as well as edge and error cases.

Please let me know if anything else is needed!

[GiovanniCanali]

Thank you, @s-chudmunge, for addressing the minor issues discussed above. Everything looks good to me!