fix(upload_large_folder): avoid infinite retries on non‑retryable commit errors

[PredictiveManish]

Details:
upload_large_folder was getting stuck in an infinite retry loop when commits failed with non retry-able errors, most notably the 403 "storage patterns tripped our internal systems" storage‑limit error. The worker logic would keep putting the same files back into the commit queue and continuously rehash/retry them. and solves #3325.

• This PR fixes Infinite retry loop in upload_large_folder when encountering HTTP 403 storage limit errors #3325
  This PR introduces a small abort mechanism in the large upload scheduler:

• LargeUploadStatus now tracks a fatal_error and an aborted flag.

• The COMMIT worker inspects HfHubHTTPError and treats the 403 storage‑limit message as a fatal error instead of something to retry.

• When such an error occurs, we mark the upload as aborted, stop re‑queuing items, and re‑raise the underlying exception once workers are done.

As a result, storage‑limit issues now fail fast with a clear error instead of looping forever, while transient errors still benefit from the existing retry/backoff behaviour.