Features TBD:

* 0.05    generate (random var with $self's distribution)
* 0.05    total_abs_error, mean_abs_error - think more
* 0.07    data inspection tools - allow user to look at real bin counts
* 0.07    framework for sophisticated algorithms over data
* 0.08    add multipoint interpolation to sum_of/mean_of
* 0.08    write script for studying how mean etc converge with smaller bases
* 0.08    add infinity abs threshold to disregard BIG values
* 0.08    add abs($x - mean)**power (who needs it?)
* 0.08    TST metatest: i.e. analyze several samples, compare results to past.
* 0.08    discarding unrepresentative data (small bin counts, far from others)
* 0.08    add antialiasing to histogram

Cleanup:

* more detailed docs on methods
* rationale

Features for discussion:

* better mode algorithm
* Additional module with smooth data model (more CPU, more precision, exact mean)
* how to find the bulk of a distribution
* C++/XS library
* analysis tools (like: Your sample looks like 45% normal (100, 15) + 35% exp (100) + 20% utter rubbish)

Multidimentional S::D::LogScale:
* linear/log storage
* correlation for 2D
* all other stuff