- OptimizedTelegramScraper accepts account_id parameter;
sets DATA_DIR to data/accounts/<id>/ when provided
- Session path uses account_session_path() for per-account session files
- initialize_client() reads per-account credentials from account state store
- ScraperJobService passes account_id from payload to scraper constructor
- Per-account state loaded from AccountStateStore before scraping
- Graceful shutdown: SIGTERM/SIGINT handlers stop continuous scraper,
disconnect Telegram client, drain running jobs with timeout
- Logging: replace print() with logging module throughout web server,
basicConfig in main.py with structured log format
- Streaming: chunked file reads (64KB) instead of loading entire files
into memory; full HTTP Range Request support (206 Partial Content)
for video seeking in the browser
- Cache: TTL-based in-memory cache (1s) for state.json reads in
StateStore, thread-safe with existing RLock
- Bugfix: send_error_json now converts HTTPStatus to int for JSON
forust