Add benchmark framework to sway-lsp

[JoshuaBatty]

Description

This PR adds a way to benchmark the LSP requests and TokenMap methods using criterion. I've also added a large sway benchmark project that can be used when running benchmarks now and in the future to measure performance improvements.

Here is the result of the benchmarks.

Request Type	duration
hover	442ms
documentHighlight	441ms
definition	433ms
prepareRename	433ms
rename	436ms
semantic_tokens	464ms
document_symbol	447ms
completion	893ms
inlay_hints	15ms
code_action	444ms
code_lens	34ms
on_enter	572ms

TokenMap method	duration
tokens_for_file	230ns 10ms
idents_at_position	438ms
tokens_at_position	437ms
token_at_position	439ms
parent_decl_at_position	437ms

Compilation	duration
compile_and_traverse	1.795sec

related to #4901

Checklist

• I have linked to any relevant issues.
• I have commented my code, particularly in hard-to-understand areas.
• I have updated the documentation where relevant (API docs, the reference, and the Sway book).
• I have added tests that prove my fix is effective or that my feature works.
• I have added (or requested a maintainer to add) the necessary Breaking* or New Feature labels where relevant.
• I have done my best to ensure that my PR adheres to the Fuel Labs Code Review Standards.
• I have requested a review from the relevant team or maintainers.

[anton-trunov]

LGTM! Perhaps we should add some kind of developer documentation where it says we actually have benchmarks and how to run those?

[xunilrj]

Some methods like tokens_for_file are lazy. Maybe that is why their result is so good. You should collect() its results to a Vec.

I am also wondering if it is worth to serialize responses to JSON. Because this way we also assess the "size" of the response.

[sdankel]

Awesome work!

If the benchmarks don't take too long to run, it would be cool to integrate the request benchmarks into CI with a script like this. We could detect any change in the compiler or LSP code that significantly slows things down.

[JoshuaBatty]

Some methods like tokens_for_file are lazy. Maybe that is why their result is so good. You should collect() its results to a Vec.

Good point, will modify the benchmark to collect into a Vec for this one.
I'll have a think about measuring the size and address that in a later PR. Thanks for the review!

[JoshuaBatty]

If the benchmarks don't take too long to run, it would be cool to integrate the request benchmarks into CI with a script like this. We could detect any change in the compiler or LSP code that significantly slows things down.

Nice, yeah something like that could be cool to integrate into our CI down the track. Will have a think about what would be useful for us.

[JoshuaBatty]

LGTM! Perhaps we should add some kind of developer documentation where it says we actually have benchmarks and how to run those?

I've added a page under the Engineering/LSP page in the internal Notion docs for now. We could perhaps add another .MD to this repo at some point if the need arrises.

If you want to run them yourself just cd into the sway-lsp directory and run cargo bench

[sdankel]