Inconsistent results for properties that introduce several error locations

[Heizmann]

Some C properties are translated into several error locations (e.g., memsafety).
If only some of the error locations are reachable and we do not stop after the first violation was found, we report that the property holds and that the property does not hold.

A less severe symptom of this problem is that we sometimes report several positive results for the same property.

A solution:
Let baktranslation merge results.

[danieldietsch]

Makes sense.

For this to work we could, e.g., mark all the asserts that belong to one property with the same check instance. I will look into it.