Refactor benchmarks

[cupicmarko]

Revamped benchmarks:

• Benchmarks no longer run any code outside of bench_function block
• Using latticefold/benches/config.toml to run benchmarks
• No macros used in benchmarks
• Using build.rs to generate benchmarks code from latticefold/benches/config.toml

[cupicmarko]

Added some comments for clarity, diff is hard to read

[v0-e]

I think it looks pretty cool! 💯
Just a small improvement: with my experience in benchmarking this repo 😄, it would be nice to have the possibility to choose which subprotocol to bench, for example, just run the folding benches. Also the same thing for scalar, non-scalar, 3 non-scalar.
You can make fields an Option in the benches config, and generate the benches based on their existence 👍

[cupicmarko]

I think it looks pretty cool! 💯 Just a small improvement: with my experience in benchmarking this repo 😄, it would be nice to have the possibility to choose which subprotocol to bench, for example, just run the folding benches. Also the same thing for scalar, non-scalar, 3 non-scalar. You can make fields an Option in the benches config, and generate the benches based on their existence 👍

I have something that can filter benchmarks based on environment variables, I just have to verify it works and I'll push one commit

[maksimryndin]

awesome job, @cupicmarko! 👍

[cupicmarko]

It was updated to allow filtering, also bounds.sage was updated based on @ivolkg changes. I've noticed that it takes significantly more time to compile and link benches because there is more benchmark tests, so for local running, I would suggest either building benches incrementally or linking dynamically, to reduce build time.