Benchmarking Vision-Language Contrastive Methods for Medical Representation Learning