Outlier Detection with the IQR Rule

The IQR rule flags values far outside the middle 50% of the data. Unlike z-scores, it isn't thrown off by the outliers themselves.

\text{IQR} = Q_3 - Q_1, \qquad \text{outlier if } x < Q_1 - k \cdot \text{IQR} \text{ or } x > Q_3 + k \cdot \text{IQR}

Implement iqr_outliers(values, k=1.5) returning (lower, upper, outlier_indices):

$Q_1$ and $Q_3$ are the 25th and 75th percentiles from np.percentile (its default linear interpolation).
outlier_indices is a sorted list of the positions of values strictly outside [lower, upper].
Ignore NaN values when computing the quartiles, and never flag them.