Power-Law Dynamics in an Auditory-Nerve Model Can Account for Neural Adaptation to Sound-Level Statistics