More future-proof type definitions (Algebraic data types)

[ekpyron]

[just dumping my comment from matrix] For the high-level type descriptions, I'm still wondering if we should make it more future-proof from the start. We will have general algebraic data types in the future and enums and structs are just special cases of those equipped with additional name information. I'm fine with just representing the current state of the language in the format, but that means we'll very likely need to break and extend this in the not-so-distant future, so I'd recommend generalizing this a bit more from the start.