Make benchmark setup consistent

[BD103]

Objective

• Benchmarks are inconsistently setup, there are several that do not compile, and several others that need to be reformatted.
• Related / precursor to Prefix benchmark names with module path #16647, this is part of my attempt migrating bevy-bencher to the official benchmarks.

Solution

Tip

I recommend reviewing this PR commit-by-commit, instead of all at once!

In 5d26f56 I reorganized how benches were registered. Now this is one [[bench]] per Bevy crate. In each crate benchmark folder, there is a main.rs that calls criterion_main!. I also disabled automatic benchmark discovery, which isn't necessarily required, but may clear up confusion with our custom setup. I also fixed a few errors that were causing the benchmarks to fail to compile.

In afc8d33 I ran rustfmt on all of the benchmarks.

In d6cdf96 I fixed all of the Clippy warnings.

In ee94d48 I fixed some of the benchmarks' usage of black_box(). I ended up opening rust-lang/rust#133942 due to this, which should help prevent this in the future.

In cbe1688 I renamed all of the ECS benchmark groups to be called benches, to be consistent with the other crate benchmarks.

In e701c21 and 8815bb7 I re-ordered some imports and module definitions, and uplifted fragmentation/mod.rs to fragementation.rs.

Finally, in b0065e0 I organized Cargo.toml and bumped Criterion to v0.5.

Testing

• cd benches && cargo clippy --benches
• cd benches && cargo fmt --all

[bushrat011899]

Looks good to me! Definitely a cleaner structure.

[MrGVSV]

I was previously running function-reflection-specific benches like:

cargo bench --bench reflect_function --all-features

Does this change mean I won't be able to run specific sets of benches anymore?

[BD103]

Yes, but only temporarily! Just like unit tests, you can filter what subset of benchmarks you want to run:

# Will run `typed/function`, `typed/closure`, `typed/closure_mut`, and more.
cargo bench --bench reflect -- typed

My plan with #16647 is to rename all benchmarks to include their module paths, so you'll be able to run:

# This is equivalent to `cargo bench --bench reflect_function`, using the new system.
cargo bench --bench reflect -- bevy_reflect::function

This PR was a precursor, since I didn't want to have too many changes at once.

[Trashtalk217]

I want to second this, because when I still was putting the stress test , it was very annoying to test just the bench while not testing everything else. Can you still run specific benchmarks (or groups of them) with this reorganization?

[Trashtalk217]

Looks good!

[alice-i-cecile]

@BD103 once this is merge conflict free I'll press the button. Thanks for this: the benchmarks folder has been in a state of disarray for way too long.

[BD103]

@alice-i-cecile should be good to go now!