Move contractions tiling logic outside of Contractions_NT

[ahendriksen]

The main functionality of Contractions_NT involves loading tiles of data into shared memory to enable fast GEMM-like kernels. In practice, this requires keeping track of tiles of data (2D submatrices of a bigger matrix) and distributing the data in the tiles over shared memory and registers of thread in a thread block.

Currently, Contractions_NT performs indexing logic for both:

1. The distribution of data in a tile over registers and shared memory;
2. Looping over tiles of data in a 2D matrix.

In this PR, we move functionality 2 out of Contractions_NT. Moving over the tiles of data and keeping track of the grid stride loop is now the responsibility of the calling code.

Splitting these responsibilities is helpful when non-trivial tiling logic is required, as in the upcoming sparseL2NN functionality.

Note: This PR also cleans up one unfortunate wart in the current implementation. Depending on which of the two overloaded constructors was called, the tiling logic was transposed leading to extremely difficult to track down bugs.

[ajschmidt8]

Please use GitHub's Draft PR feature instead of WIP tags in the future. Draft PRs have the benefit of preventing notifications to codeowners until PRs are marked Ready for Review, which helps cut down on excessive notifications for PRs that are still being worked on. CI will still run on Draft PRs.

Some useful information about Draft PRs:

• https://github.blog/2019-02-14-introducing-draft-pull-requests/
• https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/proposing-changes-to-your-work-with-pull-requests/about-pull-requests#draft-pull-requests
• https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/proposing-changes-to-your-work-with-pull-requests/changing-the-stage-of-a-pull-request

[ahendriksen]

@tfeher @cjnolet : I have rebased the branch onto the latest 23.02 branch. It would be nice to get this PR merged. Both for sparseL2NN and hopper-based pairwise distance kernels.

[codecov-commenter]

Codecov Report

Base: 87.99% // Head: 87.99% // No change to project coverage 👍

Coverage data is based on head (4947dc8) compared to base (7c12b1e).
Patch has no changes to coverable lines.

Additional details and impacted files

@@              Coverage Diff              @@
##           branch-23.02     #837   +/-   ##
=============================================
  Coverage         87.99%   87.99%           
=============================================
  Files                21       21           
  Lines               483      483           
=============================================
  Hits                425      425           
  Misses               58       58           

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[cjnolet]

Holding off on merging right away so we can investigate the CI timeouts.

[cjnolet]

/merge