chore: formatter caching

[tippenein]

simple caching to speed up benchmarks

formatter                     fastest       │ slowest       │ median        │ mean          │ samples │ iters
╰─ format_benches                           │               │               │               │         │
   ├─ format_all_contracts    34.29 ms      │ 43.32 ms      │ 34.52 ms      │ 35.38 ms      │ 10      │ 10
   ├─ format_clarity_bitcoin  8.751 ms      │ 10.11 ms      │ 9.931 ms      │ 9.846 ms      │ 10      │ 10
   ╰─ format_heavy_section    2.221 ms      │ 2.221 ms      │ 2.221 ms      │ 2.221 ms      │ 1       │ 1

Main changes:

• use slice::from_ref when passing exprs into recursion
• cache using the pointer to that expr or list of exprs (display_pse and format_source_exprs)

[codecov]

Codecov Report

Attention: Patch coverage is 96.51163% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
components/clarinet-format/src/formatter/mod.rs	96.51%	3 Missing ⚠️

📢 Thoughts on this report? Let us know!

[hugocaillard]

Awesome work on the perf 🙌

I'm wondering if the CACHEs need to be top level RefCell?
Could it be stored in the Aggregator directly?

pub struct Aggregator {
    ...,
   expr_cache: HashMap<(usize, String),
   list_cache: HashMap<(usize, String),
}

Or what are benefits with the refcell approach here?

[tippenein]

@hugocaillard simplified the cache a bit. Also removed the redundant segment of raw source from clarity-bitcoin since we already test that.

[hugocaillard]

lgtm ⚡