Allow for dynamic properties in Object

[Dinnerbone]

Nothing uses them yet, but we will need to change any versioned global method to use it.

Opening a PR before then in case anyone else wants to go make that change, and to avoid conflicts in Object

[Herschel]

Thank you! Instead of using NativeFunction, do you think we should have a type Getter = fn(...) -> Value<'gc> and type Setter = Option<fn(..., Value<'gc>)> for Property::Virtual { get, set } ?

[Dinnerbone]

I imagine it'll be Executable<'gc> once #63 is in. It will be either a native or user defined function, so the typedef would be type Getter = Executable<'gc> and type Setter = Option<Executable<'gc> - I'm not sure that'll be beneficial.

[Dinnerbone]

One other thing: I actually disagree with the clippy lint about preferring to using .. for the debug method as it's deliberate that get is ignored, and it should error if other properties are added (they should likely be printed too).

[Herschel]

I imagine it'll be Executable<'gc> once #63 is in.

So then the question is whether to add Exectuable::NativeGetter etc. Not sure it's that beneficial -- my thought is it gets rid of the bounds check when accessing the arg in a setter.

One other thing: I actually disagree with the clippy lint about preferring to using ..

Makes sense! Actually looks like this lint might be removed altogether. I might just allow this lint globally.

Merging, thank you!