Inference to the best explanation as supporting the expansion of mathematicians’ ontological commitments