Pretrained Conformers for Audio Fingerprinting and Retrieval