Show parent name in terminal output (Class::method_name)

[nminev]

This is a follow-up to #94 which was closed. Splitting that PR into two smaller ones — this covers the output display change only.

What

Nested entities now display as Parent::name in the terminal formatter:

// before
│  ↻ method     get_card_1              [renamed]

// after
│  ↻ method     CardService::get_card…  [renamed]

When a file has multiple classes with similarly-named methods, the current output gives you no way to tell which class a change belongs to without going to the file. This adds that context directly to the diff output.

Changes

• SemanticChange gets a new parent_name: Option<String> field (serialised as parentName in JSON, omitted when null — fully additive, #[non_exhaustive] already covers this)
• make_change() in identity.rs populates it from entity.parent_id
• Terminal formatter uses it to prefix the name

Also fixes a latent UTF-8 panic in the truncation path — the old code sliced by byte index which would panic on non-ASCII names. Replaced with char_indices.

Tests

Existing tests all pass. The field is exercised end-to-end through the existing formatter output.

[inspect-review]

inspect review

Triage: 5 entities analyzed | 0 critical, 0 high, 3 medium, 2 low
Verdict: standard_review

Findings (2)

1. [low] Terminal formatter truncation uses display_name.len() (byte length) to decide whether to truncate, but truncates by character index (char_indices().nth(24)). For non-ASCII names, len() > 25 can be true even when there are <= 25 characters, causing unnecessary truncation and incorrect output. Evidence: if display_name.len() > 25 { ... display_name.char_indices().nth(24) ... } in crates/sem-cli/src/formatters/terminal.rs.
2. [low] Terminal formatter truncation produces a string longer than the intended 25-character column: it takes the first 24 characters (nth(24) gives the start index of the 25th char, so slicing [..i] yields 24 chars) and then appends an ellipsis, resulting in 25 visible chars, but format!("{:<25}", truncated) will not truncate longer strings; if the slice yields 25 chars in some cases (depending on boundary), appending … can exceed 25 and break alignment. Evidence: format!("{}…", &display_name[..i]) followed by format!("{:<25}", truncated) in terminal.rs.

---

Reviewed by inspect | Entity-level triage found 0 high-risk changes

[rs545837]

Thanks for opening the PRs, sorry got a bit packed with schedule, will review this today.

[rs545837]

Thanks for the PR, it really does make sem more easier to view the diffs, and for the UTF-8 fix.