Change width and height of Size to Option<T> instead of T

[Weibye]

What problem does this solve or what need does it fill?

The geometry::Size<T> is currently defined as

pub struct Size<T> {
    /// The x extent of the rectangle
    pub width: T,
    /// The y extent of the rectangle
    pub height: T,
}

While digging through the code in relation to #144 and #148 I've started wondering if we should change the definition of Size to the following:

pub struct Size<T> {
    /// The x extent of the rectangle
    pub width: Option<T>,
    /// The y extent of the rectangle
    pub height: Option<T>,
}

The width and height fields need to be independently defined or not, which means that replacing everything with some_field: Option<Size> will not cover all use-cases throughout the code.

I'm not yet able to see the full picture how everything fits together once we:
Dimension::Undefined -> Option<Dimension>
Number::undefined -> Option<f32>

but it seems it is a must for this to work (?)

What alternative(s) have you considered?

Leave it as is, but I expect it to change in some way or another.

Are there alternative designs we can explore?

[alice-i-cecile]

This seems like about the right approach. In general it really seems like we're typically storing options here.

[alice-i-cecile]

I'm tackling this now.

[alice-i-cecile]

Actually, this appears to be blocked on #148. @Weibye, feel free to attempt this in a follow-up, or bug me to do it :)

[Weibye]

Actually, this appears to be blocked on #148. @Weibye, feel free to attempt this in a follow-up, or bug me to do it :)

First I thought we'd need so solve #148 before this, then I thought this needed to come first. I supsect they both need to be done at the same time for the whole system to work, but I'll keep tinkering with it tomorrow morning :)

[nicoburns]

This doesn't won't work now we're also using Size with AvailableSpace. And Size<Option<f32>> is working well as is :)

[Weibye]

👍 It seemed to make sense at the time :)