release: v0.25.0 — caption follows the transcript's script (CJK cues without spaces, character-bounded)

[renezander030]

What changes

• caption follows the transcript's script. Whisper's "words" for Chinese and Japanese are single characters or short tokens, so the Latin defaults (four words per cue, joined with spaces) produced fragments with spaces between the characters. Cues in Chinese and Japanese are now joined without spaces and bounded by characters alone, at the width lint holds captions to (16 zh, 13 ja; Korean keeps spaces and four words at 16). Karaoke ranges follow the new offsets. The result reports caption_script (latin | zh | ja | ko). An explicit --max-words / --max-chars applies as given.
• Library: groupWords takes a separator; wordSeparator, groupingDefaults, GroupingDefaults exported; the script detection moved to src/script.ts and stays exported from the lint entry points.
• caption --max-chars documented next to --max-words; the Chinese README notes the feature in 科技爱好者周刊 issue 413; version 0.25.0, changelog, README highlights (v0.25.0 and v0.24.0 kept).

Verification

• npm run lint clean; full suite green on the pre-commit hook; new test/caption-cjk.test.mjs (script detection, separators, grouping defaults, groupWords with and without a separator, and a fake-whisper karaoke run on an 18-character Chinese transcript: space-free cues of at most 16 characters, third word's highlight range [2, 3], --max-words 6 honoured).

Review notes

• Base: 72cd7d3 (v0.24.0). Touched: src/caption.ts, src/script.ts (new), src/lint.ts (moved definitions, re-exports), src/lib.ts, src/command-specs.ts, READMEs, changelog. Untouched: --script alignment still tokenizes the script file on whitespace (a Chinese script line is one token to it), import-srt, rendering.
• Behaviour change only for CJK transcripts; Latin output is byte-identical (existing caption tests unchanged).