Add black_box to some benchmarks that need it

[nishantc1527]

Objective

The black_bxo function in Criterion is essential to make consistent benchmarks because LLVM likes to optimize things out randomly. It's a very weird and confusing process, it takes into account the size of the overall function and some random other stuff that's impossible to predict. So even though benchmarks work now, they may not work in the future. I wasn't able to reproduce any of this in the current state, but that's because it's so hard to predict LLVM.

Fixes #7156

Solution

I looked across the benches crate and found a few places that needed black_box. I didn't do a thorough check, I just kind of skimmed the files to look for things that stood out (constant numbers are usually a good tell, as well as unused variables).

Testing

It's a very hard thing to test because it's hard to reproduce. But looking at the docs (https://doc.rust-lang.org/beta/std/hint/fn.black_box.html) it's clear that every place I put it is necessary.

---

[github-actions]

Welcome, new contributor!

Please make sure you've read our contributing guide, as well as our policy regarding AI usage, and we look forward to reviewing your pull request shortly ✨

[alice-i-cecile]

@mockersf I'm in favor of this, but giving you a heads-up because this will introduce spurious changes in the benchmarks.

[hymm]

Could you run before/after on the benchmarks that have changed? That way we can know what to expect in CI if there are any measurable changes.