Convert type_id from String to a node

[jansul]

This PR modifies the type_id/type_id! methods of the parser to return a new node (Ast::TypeId) rather than return a String. Any other node affected by this has also been updated.

This should make it easier to implement language server features where it can be useful to exact positions of some nodes. e.g. the name of an Ast::Component

I've tested this change against the mint-realworld repository, running mint docs etc and everything seemed OK, but could possible use more testing given how much is changed.

[gdotdesign]

Looks good 👍 I guess if it compiles and works with the real word repo it should be good! Once merged I'll test mint-ui with it.

[Sija]

I'm wondering, wouldn't be cleaner to implement TypeId#to_s and Comparable(String) to avoid explicit usage of #value property?

[gdotdesign]

I'm wondering, wouldn't be cleaner to implement TypeId#to_s and Comparable(String) to avoid explicit usage of #value property?

I guess we could do that, however we currently don't, so I don't know if it's in the scope of this PR or not 🤔

[Sija]

@gdotdesign I'm pretty sure we could at least implement TypeId#==(other: TypeId) to avoid doin' in manually everywhere, like:

ast.stores.find(&.name.value.==(connect.store.value))

[jansul]

I was very tempted to implement those methods, however as it's not something we are doing on any other node (especially with Ast::Variable where we also use .value) I thought best to do it like this for consistency.

Happy to make any changes 😄 TypeId#==(other: TypeId) would probably be the safest and least confusing.

Just noticed a bug where I'm not using .value when passing to errors so I will fix that too ...!

[jansul]

Fixed all of the above 😄

As an aside it may be worth hardening the bold method of BlockBuilder to only accept strings the same as text as that would have caught that issue!