Scala 3

[bblfish]

The move to Scala3 requires a very serious rewrite, as Scala 3 has dropped projection types.
I started a conversation on the lampepfl github discussion space to see how others are coping with this. That has a number of important links too. But I should perhaps move the conversation to a more visible forum.

Anyway, I think when I last tried to compile banana-rdf in Scala3 I ended up with way over 500 problems.

The obvious answer is to use projection types, but it is not that easy. I tried 10 months ago, but at the time my understanding of Scala3 was very small.

[bblfish]

Projection types are not the panacea either I think. Using bananas-rdf in a Scala3 project, I have noticed that there are two problems:

1. Inheritance of types does not work. So inference from Rdf#Bnode to Rdf#Node do not work automatically.
2. There is leakage from the underlying projection types if one does not parameterize one's code with Rdf in class SomeClass[Rdf <: RDF](...). I noticed this because I wanted to add a find method to GraphW that returns an Iterator. But if I have Rdf hard code to Jena I get

				acRulesGraph.find(node,wac.default,ANY).exists{ (cnode: Rdf#Node) =>
[error]    |					^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[error]    |value exists is not a member of org.apache.jena.util.iterator.ExtendedIterator[org.apache.jena.graph.Triple], but could be made available as an extension method.

So Scala3's opaque types may help avoid this type of problem.