Add built-in indexof/valueof types

[dcodeIO]

This PR investigates a special type valueof<T> that obtains the value type of an array, a set, a map or a class with an indexed get overload (like typed arrays).

Might help to implement things like #516 but the implementation is actually a bit short-sighted. TypeScript supports resolving types of the form T[0] for example, that does exactly this, but the resolver isn't yet powerful enough to support this (resolving of any expression etc.). So I guess the question is: Does it make sense to add this now, and once the resolver has been reworked, simply keep supporting it for backwards compatibility?

[MaxGraey]

Nice. What about keyof?
Btw idiomatic way to retrieve value in TS:

type ValueOf<T> = T[keyof T];

[dcodeIO]

That's what I meant with a bit short-sighted. A proper implementation would first need a new type node kind keyof Something, then a new type node kind T[Something] and finally the resolver must be able to resolve that. Not a 100% sure, but since the resolver requires a rework anyway (resolve any expression etc.), it looks like doing this properly would depend on doing that first. Really more a hack in this form, but maybe a useful one so we can get TypedArray#set out the door.

[MaxGraey]

I see. So as intermediate solution this definitely make sense

[jtenner]

Count me in! This enables many great comparison patterns for my testing software.

[dcodeIO]

Thinking about this a little more, I figured that having our own special types might actually better represent what these actually do, because in TS, keyof doesn't just get the indexed access key, but all possible keys

type TheType = keyof Array<i32>; // number | "length" | "fill" ...

which is most likely nothing we'd support anytime soon anyway. Now, sadly, keyof is a keyword and we can't have a keyof<T> type, but we could just name ours indexof<T> obtaining the index type and valueof<T> obtaining the value type of indexed access. Wdyt?

[jtenner]

It would be awesome to have a type declaration statement in the function body.

type Child = indexof<T>; // etc. 

Just my 2 cents. At first glance, I think this might cause problems with the resolver. I would be grateful for the type itself without the statement :) .

[dcodeIO]

This now adds indexed access overloads for Map and Set to unify/simplify the concept by binding it to [] overloads specifically, and introduces the corresponding indexof<T> type.

[dcodeIO]

Docs: https://docs.assemblyscript.org/basics/types#macro-types