Currently gives None for a 16 bit greyscale image rather than the true min and max values in the 0-65536 range. The internal ImagingGetProjection function already supports I;16 but the _getextrema needs to know to unpack the result.
