equivalenceHash makes use of hashCode

[kwalcock]

The equivalenceHash is used to generate IDs which get serialized. The value should be the same for the same documents and sentences, regardless of Java version or Scala version. In some places, "string".hashCode is used, but those values do seem to have been standardized in Java. Also used is directedGraph.equivalenceHash that calls edges.hashCode, which is not standardized. The edges are stored a fairly abstract List and the hashCode has changed between Scalas 2.11/12 and 2.13+. The code needs to be changed so that it only depends on the values and not the container.

The fix isn't scheduled yet, because it will break old hash codes and just now it is important to keep them for regression testing until it's clear that the json serialization changes are correct.