perf(transformer): common transforms store state in UnsafeCell

[overlookmotel]

Experiment. Not for merging.

As an alternative to #11742, we could wrap state in common transforms in UnsafeCell instead of RefCell. We'd need a nice way to do that (probably would have to be a proc macro, unfortunately). This PR just does it manually with an ugly method, to see what the perf gain is, and so how worthwhile it would be to do it properly.

[graphite-app]

Your org has enabled the Graphite merge queue for merging into main

Add the label “0-merge” to the PR and Graphite will automatically add it to the merge queue when it’s ready to merge. Or use the label “hotfix” to add to the merge queue as a hot fix.

You must have a Graphite account and log in to Graphite in order to use the merge queue. Sign up using this link.

[overlookmotel]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• perf(transformer): common transforms store state in UnsafeCell #6246 👈
• refactor(transformer): move functionality of common transforms into stores #6243
• refactor(transformer): rename var in VarDeclarations common transform #6242
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @overlookmotel and the rest of your teammates on Graphite

[codspeed-hq]

CodSpeed Performance Report

Merging #6246 will not alter performance

_{Comparing 10-02-perf_transformer_common_transforms_store_state_in_unsafecell_ (810cfb5) with 10-02-refactor_transformer_move_functionalty_of_common_transforms_into_stores (a0a3a8b)}

Summary

✅ 29 untouched benchmarks

[overlookmotel]

Only +0.2% perf boost at best. I'm really surprised.

Perhaps I've done it wrong, or perhaps compiler was able to deduce that RefCell::borrow_mut could never panic, and optimize out the checks, so using UnsafeCell offers little advantage.