fix: reduce chess-results.com request volume by ~20-50x

- check_all_subscriptions: quick pre-check (1 req, art=4 only) before
  full fetch; full fetch only when results_count or current_round changed.
  Interval 60s → 300s. Saves ~10,000 req/day per active subscription.
- rescan_existing_tournaments: filter to tournaments ended within last 30
  days (or no end_date). Drops from ~2000 cached TNRs to ~30-100 active
  ones per run, saving ~22,000 req/day.
- warmup: sleep 1s per TNR instead of 20ms per 500 — spreads initial
  full-scan over hours rather than minutes.
- parser: extract _normalize_url helper, add fetch_tournament_quick().

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

bots/client_bot.py

@@ -734,7 +734,7 @@ def main():
     app.add_handler(MessageHandler(filters.TEXT & ~filters.COMMAND, handle_url))
 
     app.job_queue.run_repeating(
-        tracker.check_all_subscriptions, interval=60, first=10)
+        tracker.check_all_subscriptions, interval=300, first=10)
     app.job_queue.run_repeating(
         tracker.rescan_new_tournaments, interval=3600, first=60)
     app.job_queue.run_repeating(