Task 070: 文字起こしページ複数ソース表示

Status: DONE

Motivation

Human directive: 「文字起こしのページは、複数のデータソースの生の表示と、文脈から文字や話者を修正したものをそれぞれ選択可能にしたい」

Completed

• Added additionalSources field to TranscriptionPageData type
• Updated discoverTranscriptions() to find epXX_lines_*.json alternative source files
• Copied Whisper lines data to reports/data/episodes/ep01-04_lines_whisper.json
• Implemented tab UI with up to 3 views:

- 修正版（話者帰属済み）: Phase 2 corrected dialogue with speakers

- YouTube 自動字幕（生データ）: Raw YouTube VTT extraction

- Whisper STT（生データ）: Raw Whisper STT extraction

• Tab switching via lightweight inline JavaScript
• Single-source pages render without tabs
• Source info card shows all available sources with line counts
• Transcription index shows multi-source counts
• 4 new tests (tab UI rendering, single tab, two tabs, source card)
• All 965 TS tests pass, typecheck clean