Refactor for wgpu v23 compatibility

[AsherJingkongChen]

Checklist

• Confirmed that run-checks all script has been executed.
• Made sure the book is up to date with changes in this PR.
  • After this PR merged, contributor-book/src/guides/adding-a-new-operation-to-burn.md needs to be updated.
• Check dependencies.

Related Issues/PRs

• Dependent on v23.0.0 release gfx-rs/wgpu#6465
• Blocked by Refactor for wgpu v23 compatibility with an example of wgpu device sharing cubecl#211
• Blocked by Proper bounds checking for CubeCL kernels in Burn crates #2451

Close #2451

Changes

• Update cubecl to tracel-ai/cubecl@13fc62c
• Update wgpu to 23.0.0
• Update WgpuDevice::id() in burn-tensor

Testing

(Solved): I run ./run-checks.sh all, but too many errors occurred on macOS (Metal) 🥺. Fixing the bugs ...

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 1 line in your changes missing coverage. Please review.

Project coverage is 81.86%. Comparing base (6dbc3fd) to head (9ac5a3d).

Files with missing lines	Patch %	Lines
crates/burn-tensor/src/lib.rs	0.00%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2435   +/-   ##
=======================================
  Coverage   81.86%   81.86%           
=======================================
  Files         833      833           
  Lines      106449   106448    -1     
=======================================
  Hits        87146    87146           
+ Misses      19303    19302    -1     

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

[AsherJingkongChen]

Discovery

The least recent version of wgpu that breaks burn-wgpu tests on Metal is found, see gfx-rs/wgpu@9c781d6.

[AsherJingkongChen]

Status Update

Currently, I am busy working at my personal projects. I would run CI every one or two weeks.

It seems that there are more errors occurred on macOS. I would resolve them after I finish my own works.

[ArthurBrussee]

That's awesome! Thanks for checking again :) I think the remaining errors are the same as here: #2539

The deformable convolutions might need some attention from Genna, for the others, maybe @nathanielsimard knows?

[AsherJingkongChen]

It just passed CI 20 hours ago ... And macOS failed at 110+ tests?

I think I am using the right CubeCL version.

https://github.com/tracel-ai/cubecl/compare/1c4e0036c043422773fd6824c2a888160fca8e5e..13fc62c22f311d2d153abbc4ffb4ef5271bc82a3