Define bad configurations

[Céline Bensoussan]

In the resolveConflicts of the getRealizationModels of the weaving phase when making a feature selection, it would define some configurations as being bad when the resolved conflict resolution models were realization common features.

The bad configuration cases were removed with this commit: https://bitbucket.org/mcgillram/core/commits/13b9adf

However, we need to figure out if there are other bad configuration cases or not.

[Matthias Schoettle]

I think "bad configuration" is a bad term, since from my understanding it refers to the determination of conflict resolution models, which fails when there are multiple options to choose from.

For example: There's a conflict resolution model for feature A and B, and B and C. If A, B and C are selected, it would fail, but maybe the algorithm should choose both instead.

We need to discuss whether this (bad configurations) is something the weaver needs to detect. Or if this has to happen before hand.