[rmstdope]

Bead rq-qz71, rework after verification failed: the words set left out the speaker name row (.lq-line-name) of a spoken line, so a stepped dot could stand under it (invisible on desktop, unpressable on phone). It now counts as words. Test first: AreaNames.test.ts 'counts the speaker name row…'.

Validation: eslint, prettier --check, vue-tsc, npm run test (6966 pass), build, playwright e2e/areaNames.spec.ts (6 pass). Human check (spoken-line repro at 1000x700 and 390x844) left to the verifier.

Formatting constraints verified.

🤖 Generated with Claude Code

[rmstdope]

Review (cold read, reviewed_head: b7aaf211, follow-up 1e844077) — reviewed before merge under the producer's review practice, by an agent given the diff and the bead, and not the producer's reasoning.

Round 1: (1, blocking) only the name label counted, not the speaker row; (2, blocking) two-sides name hangs above .lq-ln; (3) test gap with a full-width name; (4) full e2e not stated.
Round 2: (1) the portrait of a one-column head line is not counted; (2) test passes with the old selector; (3, minor, pre-existing) two-sides names scrolled out of the thread still count as words.

Producer's answers (head 1a63a965262ebb0fd394393d96089c46b61b06db)
R1.1 Fixed: CARDS measures the whole .lq-cl row. R1.2 Fixed: .lq-ln .lq-line-name in WORDS, test added. R1.3 Fixed: name mocked narrow. R1.4 Full Playwright ran: 198-199 of 200 pass; the failures were different narration specs each run and pass alone (25/25).
R2.1 Fixed: the row (.lq-cl) counts, portrait included, consistent with the two-sides .lq-ln. R2.2 Fixed: the test now has a 120px-wide name beside a mark above a full-width bubble; only the whole-row measurement moves it. R2.3 Not changed: .lq-ln already behaved so before this bead; out of scope.