The functions erf and erfc computed with arbitrary precision and explicit error bounds (GSoc?)

The functions erf and erfc computed with arbitrary precision and explicit error bounds
S.Chevillard1
Information and Computation
Volume 216, July 2012, Pages 72-95

https://doi.org/10.1016/j.ic.2011.09.001

https://www.sciencedirect.com/science/article/pii/S0890540112000697

This is a placeholder issue identifying one implementation that would improve calculations. I cam across this in a related search, so am not in a position to implement this now.
Maybe suited to GSoC or some such project.

As @StefanKarpinski notes in issue #109 there are three use cases:

- Calculation time not an issue and default settings suffice
- Increased calculation time is preferred for more accuracy
- Decreased calculation time is preferred for less accuracy