refactor(line_index): use SIMD intrinsics for better performance

[dyc3]

Summary

This essentially uses SIMD intrinsics to add a ASCII fast path to the scanning that LineIndex does. On my machine, the benchmarks indicate speedups from ~1.5 GB/s to anywhere between ~2.8-10+ GB/s throughput. Notably, there is a regression in speed when (approximately 0.33x) for text that is made of majority unicode text. For most code (particularly JS, CSS, GraphQL), this is likely not going to be a problem, but HTML and JSON documents are more likely to have problems because those usually contain data that is meant for humans to read.

I used gpt 5.5 to do the initial research and the implementations.

Test Plan

Tested locally on my x64 machine

Using CI to test the ARM path since our runners are ARM based.

Docs

[changeset-bot]

⚠️ No Changeset found

Latest commit: 6a95add

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[codspeed-hq]

Merging this PR will not alter performance

✅ 249 untouched benchmarks

---

_{Comparing dyc3/simd-line-index (6a95add) with main (91ed677)}