fix for "Jupyter codegen: attempt to inherit from data class"

[Jolanrensen]

#178

Basically, I added a filter for data classes when passing knownMarkers on to codeGenerator.generate(), so it cannot appear as supertype anymore.

Small cleanups

[koperagen]

Soo... how did i missed that part of the PR? :D I'm not sure if val properties should be isOpen = false, unless there is a good reason, because previously schema processor would reuse existing markers from time to time thus reducing amount of generated code.

[Jolanrensen]

Well, before, I think, all were isOpen = false and it just wouldn't check it. That's where the bug came from. But other than that, the call in ReplCodeGeneratorImpl.kt:82: return generate(schema = targetSchema, name = markerInterfacePrefix, isOpen = isMutable) has been there always. Where isMutable is defined by whether the variable is a var (KMutableProperty). But I agree, it probably shouldn't matter whether it's a var or val