The same goes for planar and spherical mirrors as well, of course. 1/4 wave is pretty rough.
Generally, it's not important if a microscopic area of a mirror deviates farther from the optimal shape. For that reason, you don't specify the maximum deviation but average the error over the entire...