The problem with the high distortion turned out to be a problem between my ears. I needed to reduce the time step size in the simulation. Once I did that, I found that the distortion is negligible. When I do the FFT, an artifact of the turn-on transient is what sets the noise floor of the plot and all of the harmonics are below that. From what I can tell, the distortion has to be below 0.05%, but I don't yet know how far below. Whatever the case, it's clear that the microphone distortion will be dominated by the response of the mic element itself and not by the amplifier.