add: ROC or AUROC determination

[kadenmarika]

Quite often beside the accuracy other evaluation methods are needed, like

the ROC-curve or then the Area-under-the-ROC-curve (AUROC) value.

Problems:

• only in two class problems
• need an parameter to adjust ==> suggestion a theta for the classifier function mu (theat \in [-1 1])

If someone is intrested to integrate it, I could join and explain it in more detail :)

[EAR-Development]

Can we follow this paper for implementation details:

https://link.springer.com/article/10.1007/s00180-016-0678-y