Handle assertions and properties in subnodes

[agacek]

JKind should do something with assertions and properties in subnodes. There are different possible semantics for this, but one thing might be to just treat them all as proposed invariants.

[agacek]

I no longer like the idea of using them as proposed invariants. That's too wishy washy. Instead the assertions and properties of a subnode should be treated as full fledged properties (assuming we can figure out reasonable naming). We may want to include an option to not inline a node and instead rely on its properties.