Re: [Numpy-discussion] NA, and replacement or reimplimentation of np.ma