Fix review follow-ups for Rust APIs and benches

[zxch3n]

Summary

• return None instead of panicking in get_editor_at_unicode_pos, and add a regression test
• update stale benchmark call sites to the current text txn API and fix a pending benchmark bug
• make the trusted kv-store lazy decode path explicit and remove the extra re-boxing in diff calculators

Verification

• cargo test -p loro get_editor -- --nocapture
• cargo check -p loro-internal --benches --all-features
• cargo check -p loro-kv-store
• cargo clippy --workspace --all-targets --all-features --message-format=short

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c3fb676b97

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Keep Block::decode public for downstream callers

loro-kv-store still exposes crates/kv-store/src/lib.rs's pub mod block, so turning Block::decode into the crate-private decode_verified is a semver-breaking API regression for external crates that currently decode already-validated blocks themselves. The workspace still compiles because every in-tree caller is internal, but any downstream Block::decode(...) call will stop compiling after this patch.

Useful? React with 👍 / 👎.