Serialization normalizations

[madig]

Move more closely to ufonormalizer output.

I'm a bit mystified by our custom lib writing code. Why does the lib section end up unordered? It should be sorted alphabetically and I thought that's what plist was doing.

Closes #125
Closes #105

[cmyr]

oh no! I hadn't noticed this 😆

[cmyr]

Cool, this all looks sensible. Maybe take a look at my patches and see if there's anything interesting in there, to you?

I'll take a look at this lib-ordering business. Do you know if it's recursive? That is, if the lib contains nested dictionaries, is everything sorted during deserialization?

[madig]

I'll take a look at this lib-ordering business. Do you know if it's recursive? That is, if the lib contains nested dictionaries, is everything sorted during deserialization?

I think dictionary keys are always sorted inside a lib 🤔

[madig]

Cool, this all looks sensible. Maybe take a look at my patches and see if there's anything interesting in there, to you?

I copied your is_normal approach to advance widths. The tests I'll leave for now I think, once the lib order stuff is taken care of, my glif roundtrip test should test the exact order and formatting of all glif elements and attributes.

[github-actions]

🗜 Bloat check ⚖️

Comparing d80653f against ff64998

target	old size	new size	difference
target/release/examples/open_ufo	1.71 MB	1.71 MB	---
target/debug/examples/open_ufo	7.11 MB	7.11 MB	32 Bytes (0.00%)

[cmyr]

Cool, yea, let's keep the test because I will work on getting it passing.