Expose core unicode display-width to plugins (charWidth/stringWidth)

[Agrejus]

What this adds

A single source of truth for terminal display width and two plugin APIs built on it:

• fresh-core::display_width::{char_width, str_width} — backed by the unicode-width crate.
• editor.charWidth(codePoint: number): number — width of one code point (0 for control/zero-width, 2 for CJK/fullwidth and most emoji, else 1).
• editor.stringWidth(text: string): number — width of a whole string.

Why

This directly addresses @sinelaw's review note on #2325:

"This should be replaced by a core function exposed to plugins, and since we already do unicode width in the rust core, the implementation should use the same underlying unicode logic we have in the rust core instead of duplicating it."

Plugins otherwise have to re-derive their own Unicode width tables (e.g. markdown_compose's hand-rolled charDisplayWidth), which inevitably drift from how the editor actually measures cells. With these APIs a plugin measures width with the exact same logic the renderer uses for layout and cursor positioning, so column widths, wrapping, and alignment agree by construction.

Design

• fresh-core::display_width is now canonical. fresh-editor's primitives::display_width re-exports char_width/str_width from it (its byte↔column helpers and the DisplayWidth trait are unchanged), so there is one implementation rather than two copies.
• The width crate had to live in fresh-core (not fresh-editor) because the plugin runtime — fresh-plugin-runtime, which fresh-editor depends on, not the reverse — needs to call it. The runtime invokes the core function directly (pure, no command round-trip), matching the existing synchronous backend methods.
• fresh.d.ts updated with the two methods; ts_export API-surface list updated. (Only the two additions are in fresh.d.ts — I deliberately avoided a full regenerate, which would have pulled in unrelated pre-existing drift between the committed file and the generator.)

Risk

Low. Purely additive APIs; the editor's own width behavior is unchanged (the re-export is the identical unicode-width call it already made).

Tests / CI (all green locally)

fmt, clippy --all-features --all-targets, doc (nightly), schema diff, check --no-default-features --features runtime, plus unit tests for the new core module, the editor re-export, and the ts_export API-present / generated-.d.ts-validates tests (now covering charWidth/stringWidth).

---

Split out of #2325 (closed) as one of the non-graphics pieces. Independent of the other splits — branches from master. The follow-up markdown-compose PR will consume these to delete its duplicated charDisplayWidth table.