Compute the maximum value of a double-precision floating-point ndarray according to a mask, ignoring `NaN` values. - View it on GitHub