ELO scraper: discover the embed link and retry transient failures

Read the current OneDrive link from the stat-check.com/elo iframe on
each run, falling back to EMBED_URL if the page can't be read or its
link 404s. Retry 429/5xx and network errors up to three times with
backoff, after a one-off 503 from the Badger token endpoint failed a run.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

sites/40k-rankings/statcheck-elo-scraper/refresh_elo.py

@@ -11,7 +11,7 @@ from datetime import datetime, timezone
 from pathlib import Path
 
 from download_and_parse import download, xlsx_to_csv
-from pure_python_elo_scrape import EMBED_URL, download_url_for_embed
+from pure_python_elo_scrape import resolve_download_url
 
 HERE = Path(__file__).parent
 ROOT = HERE.parent
@@ -46,7 +46,7 @@ def main() -> None:
     json_path = Path(sys.argv[3]) if len(sys.argv) > 3 else ROOT / "public" / "elo-data.json"
 
     print("Resolving OneDrive download URL…")
-    url = download_url_for_embed(EMBED_URL)
+    url = resolve_download_url()
 
     print("Downloading XLSX…")
     download(url, xlsx_path)