Move TikTok scraper to standalone API with multi-service architecture

[karilaa-dev]

Summary

Closes #134

• Extract TikTok scraper into a standalone FastAPI service under tt-scrap/
• Introduce a service registry pattern so adding new platform scrapers (Instagram, YouTube, etc.) requires only creating a new services/<name>/ directory — no changes to core app code
• Routes are namespaced per service: GET /tiktok/video, GET /tiktok/music, etc.
• Generic exception hierarchy (ScraperError, ContentDeletedError, etc.) shared across all services
• TikTok-specific env vars now use TIKTOK_ prefix (e.g., TIKTOK_URL_RESOLVE_MAX_RETRIES)

Architecture

tt-scrap/app/
├── app.py              # FastAPI app, service registry, generic error handler
├── base_client.py      # BaseClient Protocol (contract for all services)
├── registry.py         # ServiceRegistry + ServiceEntry
├── exceptions.py       # Generic scraper exceptions
├── models.py           # Shared response models
├── proxy_manager.py    # Shared proxy rotation
├── routes/health.py    # GET /health
└── services/
    └── tiktok/         # Self-contained TikTok service
        ├── client.py   # TikTok extraction via yt-dlp
        ├── config.py   # TIKTOK_* env vars
        ├── parser.py   # TikTok response → VideoResponse/MusicResponse
        └── routes.py   # GET /tiktok/video, GET /tiktok/music

Test plan

• uvicorn app.app:app starts without errors
• GET /health → {"status": "ok"}
• GET /tiktok/video?url=<tiktok_url> → video/slideshow response
• GET /tiktok/music?video_id=<id> → music response
• GET /docs → Swagger UI shows /tiktok/video, /tiktok/music, /health
• Env vars TIKTOK_URL_RESOLVE_MAX_RETRIES and TIKTOK_VIDEO_INFO_MAX_RETRIES respected
• Old endpoints (/video, /music) no longer exist (404)