Perf/fusion deadcode

[nathanielsimard]

Added the Drop operation that can be fused, eliminating some writes in fused kernels, properly propagating dead code elimination. We could apply a dead code elimination algorithm inside of burn-fusion as well in the future to reduce even more computations that are not necessary, but this is a good start.

Uses CubeCL : tracel-ai/cubecl#712

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.30%. Comparing base (c9dce9b) to head (97128cd).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3263      +/-   ##
==========================================
- Coverage   82.33%   82.30%   -0.03%     
==========================================
  Files         978      978              
  Lines      124869   124870       +1     
==========================================
- Hits       102808   102776      -32     
- Misses      22061    22094      +33     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[laggui]

Only a minor comment, otherwise it looks pretty good!

Makes more sense now that we don't have to handle orphans across different backends / extensions 😅