POC: simple dataflow analysis

[alexanderpann]

The implementation supports detecting:

• unused assignments
• unreachable nodes (only for FailExpr at the moment)
• unused arguments (interface: IArgument)

One implementation challenge I faced is unwanted dependencies between languages that shouldn't have dependencies, such as org.iets3.core.expr.base and another language. The data flow declarations sometimes need knowledge of concepts of other languages.

Bugs /open questions:

• The data flow of fail expressions should be ignored in certain cases such as assert statements.
• Should unreachable node errors be reduced to warnings in case of false positives? Does this analysis make sense at all?
• What is the correct ancestor to start with when reporting unreachable odes?
• What is the performance limit of the analysis? I've used the Base Language limitation of 400 data flow instructions which seems quite arbitrary to me.
• How to deal with top-level content where the order of usage is not important? Should the analysis be restricted to (block) expressions?