Fold LayerInfo into Layer

[madig]

I don't think they need to be separate.

I'm not entirely sure about Layer::new_named taking a layer and dir name, but I suppose that's hard to avoid until we have a dedicated LayerSet stuct attached to Ufo that is essentially a HashMap and manages name and path name (collisions).

[github-actions]

🗜 Bloat check ⚖️

Comparing 96d24d5 against 4d9805d

target	old size	new size	difference
target/release/examples/open_ufo	1.77 MB	1.77 MB	-72 Bytes (-0.00%)
target/debug/examples/open_ufo	7.15 MB	7.15 MB	1.55 KB (0.02%)

[cmyr]

I'm trying to recall the rationale for keeping these separate. I think the idea is that a Layer is really just a bag of glyphs in some directory, but the name and path of that directory live outside of the layer, in layercontents.plist?

I don't have strong opinions here, but looking through the code it isn't clear that this patch improves ergonomics in a meaningful way? Having one less type is nice.

Maybe one way to think about LayerInfo is that our Vec<LayerInfo> is our LayerSet?

[madig]

Maybe one way to think about LayerInfo is that our Vec is our LayerSet?

I think: yes. However, not quite, as e.g. ufoLib2's LayerSet prevents deletion of the default layer, making get_default_layer never fail.

So maybe, instead, I should turn LayerInfo into a

struct LayerSet {
    contents: HashMap<LayerName, OsPath>,
    layers: HashMap<LayerName, Layer>
}

?

[cmyr]

I'm starting work today on the python bindings project, so maybe let's leave this for now and then when I go to implement the python LayerSet I'll take a look then and try to figure out a solution?