Distribution spread and location metrics using entropic separation

- New metrics for spread and location are established, applicable to any distributions.
- The measures utilise a separation distance between observations, as points of binary entropic concentration.
- Partition entropy resolves limitations of traditional differential entropy.

Distribution spread as a separation width is formalised using partition entropy, resulting in a metric for dispersion with associated end points. The binary framework resolves interpretive and other limitations of the traditional differential entropy metric, and is robust to distributional shape.