Robust z-Scores with the MAD

The mean and standard deviation are themselves pulled by outliers. The modified z-score (Iglewicz & Hoaglin) uses the median and the median absolute deviation:

M_i = \frac{0.6745\,(x_i - \tilde x)}{\text{MAD}},\qquad \text{MAD} = \text{median}(|x_i - \tilde x|)

Implement modified_z(x) and robust_outliers(x, threshold=3.5) (flags $|M_i| >$ threshold). If MAD is 0, every point that differs from the median gets $M = \pm\infty$ and points equal to it get 0.