Use xxh3.HashString128 instead of converting to byte slice

[jakebailey]

The Go compiler isn't smart enough to elide the []byte conversion from any of these. Thankfully. the xxhash library has a string variant!

Reduces the memory allocated in fourslash by 50% and makes them about 30% faster.

[Copilot]

Pull request overview

This PR optimizes hash computation by using xxh3.HashString128() directly on strings instead of converting strings to byte slices first with xxh3.Hash128([]byte(content)). This optimization eliminates unnecessary allocations, resulting in 50% reduction in memory usage and 30% faster performance in fourslash tests.

Key Changes:

• Replaced 6 instances of xxh3.Hash128([]byte(content)) with xxh3.HashString128(content) across 3 files
• No behavioral changes, as both functions return the same xxh3.Uint128 type

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
internal/project/snapshotfs.go	Updated hash computation when reloading disk file content
internal/project/overlayfs.go	Updated hash computation in 4 locations: disk file creation, overlay creation, disk text comparison, and overlay content changes
internal/execute/incremental/snapshot.go	Updated hash computation in the ComputeHash function

---

The changes look excellent! This is a clean performance optimization that:

• ✅ Consistently replaces all instances of the pattern
• ✅ Maintains semantic equivalence (both functions return xxh3.Uint128)
• ✅ Eliminates unnecessary string-to-byte-slice conversions
• ✅ Achieves significant performance improvements as stated in the PR description

No issues were found during the review. The changes are straightforward, consistent, and correct.