Bounds for stop-loss premium under restrictions on I-divergence