Update MLIR package with fix for performance improvement of one-shot-bufferize

[guillon]

Motivation

Update dependencies to last xtc-mlir packages which include the performance improvement for one-shot-bufferize.

The slow implementation of bufferization was oberved by @liamsemeria related to the PR #45.

Actually a fix was contributed to upstream MLIR with this PR llvm/llvm-project#189895

We integrated the modification as a patch on the currently packaged xtc-mlir-wheels on top of LLVM/MLIR revision 21.1.2: xtc-tools/xtc-mlir-wheels@0471d95

Description

Simply update package revisions to the corresponding xtc-mlir pypi package versions.

Commits

Note that there is also an unrelated commit to add the covergae badge to the XTC Readme.

Discussion

To be validated in term of performance improvement in PR #45

TODO

• Validate performance improvement on top of PR Mlir Backend: Tensor Dialect Support #45

[guillon]

@liamsemeria Can you confirm the expected performance improvement when cherry-picking commit "mlir: update mlir packages to to 21.1.2.8" within your branch?

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[liamsemeria]

@liamsemeria Can you confirm the expected performance improvement when cherry-picking commit "mlir: update mlir packages to to 21.1.2.8" within your branch?
@guillon with the new commit loop-explore with the tensor dialect takes around 1.5x longer than memref and before it was around 6x longer so I think the patch is working. I was expecting it to be a bit faster but its still good.

[guillon]

@liamsemeria Can you confirm the expected performance improvement when cherry-picking commit "mlir: update mlir packages to to 21.1.2.8" within your branch?
@guillon with the new commit loop-explore with the tensor dialect takes around 1.5x longer than memref and before it was around 6x longer so I think the patch is working. I was expecting it to be a bit faster but its still good.

ok thansk for the check