Fix bug with log level

The intent here was to not notify Sentry until retries were exhausted,
but logs at the "error" level triggered Sentry exceptions anyway.
Downgrade the log level to Warning resolves that.

Fixes: Sentry#THUNDERBIRD-ACCOUNTS-7Z

Author: markstos

src/thunderbird_accounts/telemetry/tasks.py

@@ -19,7 +19,7 @@ def _retry_or_report(task, exc, message):
     Keeps transient PostHog/DB/Redis outages from spamming Sentry with one event
     per retry attempt while still surfacing persistent failures.
     """
-    logger.error(f'{message}: {exc}')
+    logger.warning(f'{message}: {exc}')
     if task.request.retries >= task.max_retries:
         sentry_sdk.capture_exception(exc)
     raise task.retry(exc=exc)