Simplify Tensor::unsqueeze_dims(); improve handling of negative indicies.

[crutcher]

Checklist

• Confirmed that run-checks all script has been executed.
• Made sure the book is up to date with changes in this PR.

Related Issues/PRs

f8c845d

Changes

Independently re-discovered the unsqueeze bug fixed on HEAD at: f8c845d

This PR makes the tests more aggressive for negative offsets greater than -1; and accumulation of negative offsets.

[crutcher]

failures:

---- tests::squeeze::tests::should_unsqueeze_dims_complex_negatives stdout ----
thread 'tests::squeeze::tests::should_unsqueeze_dims_complex_negatives' panicked at /home/crutcher/git/burn/burn/crates/burn-tensor/src/tensor/api/base.rs:779:26:
source slice length (2) does not match destination slice length (1)

---- tests::squeeze::tests::should_unsqueeze_dims_negative_two stdout ----
thread 'tests::squeeze::tests::should_unsqueeze_dims_negative_two' panicked at /home/crutcher/git/burn/burn/crates/burn-tensor/src/tensor/api/base.rs:781:44:
source slice length (3) does not match destination slice length (4)


failures:
    tests::squeeze::tests::should_unsqueeze_dims_complex_negatives
    tests::squeeze::tests::should_unsqueeze_dims_negative_two

[crutcher]

There was a previous PR to try and address this, but it didn't handle -2 and other negatives correctly: f8c845d

[crutcher]

@laggui this hits your previous PR; I'll try and work a fix up later tonight.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 81.86%. Comparing base (3dc4b43) to head (047bac3).
Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2572      +/-   ##
==========================================
- Coverage   81.86%   81.86%   -0.01%     
==========================================
  Files         833      833              
  Lines      106465   106459       -6     
==========================================
- Hits        87162    87156       -6     
  Misses      19303    19303              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[crutcher]

I recognize that my limited understanding of the api means that the previous versions of the test were broken (setting the wrong target dim count); but they still showed a problem when that was fixed.

This PR works cleanly with the new test; and the simplified logic has fewer cases