Add LocaleData parameter for word/sentence segmenter

[makotokato]

Add LocaleData parameter for word/sentence segmenter

This is a part of #3284.

ICU4C has some language break rules for word and sentence segmenter, so this fix adds some rules to ICU4X per locale.

This adds LocaleData argument to all constructors. Also, locale difference is small and 2 data only, I add the override table data marker for machine state property.

[robertbastian]

praise: these sizes look much better :)

[sffc]

Nice, I approve of this architecturally, despite the extra branch at runtime. I think this is the right data model, and if it turns out to be a performance issue, we can tweak the runtime algorithm a bit.

[sffc]

Nit: This function signature is still not what we agreed in the issue, but it can be fixed later (but before 2.0)

[sffc]

Same here: please just try to load from data.

It would be okay to skip und, but please note that in the issue we agreed that the function should take an optional locale, and you should skip loading when the locale is None.

[sffc]

Nit: Please just always check the data provider. Don't rely on a hard-coded list of locales.

[sffc]

My comments were not fixed.

[sffc]

Example code for how to optionally load the locale data:

if let Some(locale) = options.content_locale {
    let request = // ...
    match provider.load(request) {
        Ok(response) => {
            self.locale_specific_data = response.payload;
        }
        Err(DataError {
            kind: DataErrorKind::IdentifierNotFound,
            ..
        }) => {
            // fall through
        }
        Err(e) => return Err(e),
    }
}

[makotokato]

My understand is that current macro (gen_any_buffer_data_constructors) doesn't support optional locale parameter.

try_new_unstable<D>(provider: &D) should be try_new_unstalbe<D>((provider: &D, Option<LocaleData>& locale), add try_new_for_locale_unstable(...) (used by datetime), add try_new_with_option_unstable(...), or something?

[makotokato]

#3284 (comment) explains about try_new_with_options_unstable

[sffc]

Nit: Please just always check the data provider. Don't rely on a hard-coded list of locales.

[sffc]

Good work!

[aethanyc]

It seems all the previous review comments have been addressed.