Thank you for your feedback! 
glad you find it interesting.
ABC methods most of the time suffer from having wider CI’s than exact methods, since both the imperfect distance metric, and the non-zero tolerance tend to affect CI’s. i think the example can be improved by considering a Kolmogorov-Smirnov distance and a lower tolerance.
by the way the example has been improved a tiny bit in the most recent docs, the most notable difference is the use of abs instead of abs2.