feat: support rspack wasi with node.js

[CPunisher]

Summary

Here are some final works to make rspack wasi available in node.js. Now you can run rspack wasi on rspack/rolldown benchmark (remember to use NAPI_RS_FORCE_WASI=1):

1. Fix regex passing from js to rs. Although I don't know why globalThis.Object.prototype is an External other than JsObject.
2. Copy canonicalize from rspack_resolver because fs::canonicalize doesn't support wasm.
3. Enable tokio multi thread feature to avoid tokio::spawn blocking the program.

BUT PAY ATTENTION:

1. Only release mode is avaliable. If you build rspack with build:cli:dev:wasm, you will get random memory errors. Currently I don't figure it out yet.
2. Rspack doesn't exit when it finishes works. Currently I don't call shutdownAsyncRuntime in js side because rayon thread pool also blocks the program. I'd like to handle these two cases in next pr.
3. You may get memory allocation failure in very big project.

Checklist

• Tests updated (or not required).
• Documentation updated (or not required).

[netlify]

✅ Deploy Preview for rspack canceled.

Built without sensitive environment variables

Name	Link
🔨 Latest commit	1350ac6
🔍 Latest deploy log	https://app.netlify.com/sites/rspack/deploys/67ebe3dc9c3a2d00099a1ba5

[codspeed-hq]

CodSpeed Performance Report

Merging #9872 will not alter performance

_{Comparing CPunisher:03-31-feat/wasi-execution (1350ac6) with main (93af8c8)}

Summary

✅ 8 untouched benchmarks

[h-a-n-a]

Thanks!

[hardfist]

is this wasi only and should be put into cfg_if!?

[CPunisher]

Here is the cfg for the impl: https://github.com/web-infra-dev/rspack/pull/9872/files/1350ac606b87cab251a08c0d77780d130674ed4f#diff-98c48758ea9967701abd9a996d1ba7ae2a0fb3afbade319da1f628f457332aa4R209

[hardfist]

why this changed?

[CPunisher]

I'm not sure but from my experiments:

1. .get_named_property_unchecked::<JsObject>("prototype") returns an External in wasm rather than JsObject.
2. But _unchecked in .get_named_property_unchecked::<Function>("toString")? is actually needless.