Add support for various ICU features

[thomasp85]

Thanks for your work on this. As I work extensively with text rendering I know how gnarly all of this is.

For my work on https://github.com/r-lib/textshaping I need access to a few ICU features that aren't currently exposed by stringi and I'd rather not have to link to ICU myself, so I was wondering if you were open to adding these to stringi?

I'm currently looking especially to get the script associated with a given unicode point, whether the glyph/character is considered full-width, as well as formatting numbers in different scripts (e.g. 45 -> 四十五).

Is such functionality something you'd consider adding?

[gagolews]

Naturally; can you identify/point out the ICU services that would be responsible for that?
Or come up with a prototype?

[gagolews]

Related feature proposals that I had for a very long time: #4, #5,

[thomasp85]

For the numeral translation I think an interface to https://unicode-org.github.io/icu-docs/apidoc/released/icu4c/classicu_1_1number_1_1SimpleNumberFormatter.html would be sufficient. It would be required to modify the locale to use a numbering system based on the users request (https://unicode-org.github.io/icu-docs/apidoc/released/icu4c/classicu_1_1NumberingSystem.html)

For my other requests it seems very aligned with #4 and #5, basically getting access to various properties of the unicode code point.

I'd be happy to try and bash some rocks together and create a PR

[gagolews]

That'd be great, thanks.

[thomasp85]

Do you know why devtools::load_all() fails with stringi, while it still builds fine the standard way? It is potentially a "me" problem but wondered if you had seen it before and knew of a fix. What I'm getting is:

> devtools::load_all()
ℹ Loading stringi
Error in `dyn.load()`:
! ICU init failed: U_FILE_ACCESS_ERROR

After everything compiles fine

[gagolews]

I've no idea...