Add Collation-Based String Sorting

[Drodt]

Previously, Hayagriva did not compare UTF-8 strings correctly #193. This is due to how Rust compares UTF-8 Strings. This PR addresses the issue by using icu_ucol for comparison, using the current locale.

Fixes #193 and adds one test to citeproc-pass.

Adds rust_icu_ucol as a dependency. Problem: icu must be installed on the user's system. How do we handle this?

I might have missed some comparison in csl_cmp.

[Drodt]

BTW: This is only a draft because I don't think it should be merged without discussing the added dependency.

Any opinion on this, @reknih @PgBiel?

[PgBiel]

Problem: icu must be installed on the user's system. How do we handle this?

It's a non-starter for us. It won't be installed on the Typst web app for example.

icu4x¹ is an alternative. However, there are concerns about the size of the collation data it pulls, and how this could impact the size of e.g. the Typst binary, Typst on the web app etc.

One alternative that was considered is to delegate this job to a trait which implements localized sorting and let upstream users (i.e. Typst) handle it. We're not fully decided on this, but it's at least better than the status quo.

Footnotes

1. https://github.com/unicode-org/icu4x ↩

[Drodt]

For now I switched to icu4x. Let me know if the typst team wants another solution