Verify that nested closure classes work properly

[zspitz]

I saw a compiler generated expression where the closure class instance was itself an instance of another closure class:

LambdaExpression
  - Body: MemberAccessExpression
      - Expression: MemberAccessExpression - <instance of closure class>
          - Constant: <instance of second closure class>

The code formatters have been fixed to treat members of an expression whose type is a closure classes as simple variables, even when the instance of said member is itself only the member of another closure class.

Nevertheless, I am unable to reproduce in tests, so I am leaving this open for now.