Revert "chore: bump noir (#12760)"

[TomAFrench]

This reverts commit e9526cf.

Please read contributing guidelines and remove this line.

[netlify]

❌ Deploy Preview for barretenberg failed. Why did it fail? →

Name	Link
🔨 Latest commit	1720b29
🔍 Latest deploy log	https://app.netlify.com/sites/barretenberg/deploys/67d968812316a3000871eb52

[TomAFrench]

cc @aakoshh I can bootstrap fine with this, but feel free to revert if you're seeing failures in master.

[aakoshh]

@TomAFrench the reason I suspect it still doesn't work is because it built so quickly:

We know that caching is not based on the actual code content until #12825 is merged, so most likely it downloaded some cached version that only hashed the contents of noir, but not noir/noir-repo. Last Friday this manifested itself later in some weird TypeScript build failure further down the line.

[aakoshh]

This is the example of a test failure: https://github.com/AztecProtocol/aztec-packages/actions/runs/13924314151/job/38964749855?pr=12825

It's using the old nargo to compile newer tests.

[aakoshh]

I'm trying to fix that in my PR, it looks like one instance of .rebuild_patterns was missed yesterday, even though I remember fixing it. If my PR builds then it should be okay.

[TomAFrench]

We should be good to close this now #12825 is merged right @aakoshh ?

[aakoshh]

It still fails on amd64: https://github.com/AztecProtocol/aztec-packages/actions/runs/13926227353/job/38971340616

But Adam thinks it's because the gh-bench command needs a git username and password specified to calculate the hash. He'll fix that, and then hopefully that's it 🤞