Add support for Croatian

[Jetman80]

This commit adds support for the Croatian language. Implementation includes cardinal numbers, EUR, USD, HRK currencies, tests. It is based on the Serbian language implementation as the closest one.

Changes proposed in this pull request:

• Croatian language support (num2words/lang_HR.py)
• Tests (tests/test_hr.py)
• __init__.py and README.rst updates to reflect the changes

Status

• READY
• HOLD
• WIP (Work-In-Progress)

How to verify this change

Run the tests for Croatian language with python3 -m pytest tests/test_hy.py -v to run tests. You can also manually test with examples like:

from num2words import num2words
print(num2words(42, lang='hr')) # četrdeset dva
print(num2words(1.5, lang='hr', to='currency', currency='EUR')) # jedan euro, pedeset centi
print(num2words(2020, lang='hr', to='year')) # dvije tisuće dvadeset

[tihomirjauk]

@Jetman80, thanks for putting this together — Croatian support has been a long gap in num2words, and the cardinal + currency implementation here is solid. I tested it against real-world Croatian text (TTS pre-normalisation for narration) and the cardinals match what a native speaker would say.

Two pieces are still missing for the PR to fully cover Croatian usage:

to_ordinal() raises NotImplementedError. Croatian ordinals are essential — “17. rujna”, “21. stoljeću”, “1986. godine” all involve ordinal forms.
to_year() falls through to to_cardinal() via the base class default, so num2words(1986, lang='hr', to='year') returns "jedna tisuća devetsto osamdeset šest" — that’s grammatically correct as a count but not how a Croatian speaker reads “1986. godine”. The natural form is "tisuću devetsto osamdeset šeste" (feminine genitive ending on the last word, “jedna tisuća” collapsed to “tisuću”).
I have a follow-up branch ready that adds both, on top of this PR’s commit:

Masculine-nominative ordinals via last-cardinal-word substitution (prvi, drugi, treći, četvrti, peti, …, dvadeseti, stoti, tisućiti)
to_year() that collapses jedna tisuća → tisuću for 1000-1999 and applies feminine-genitive endings to the last word
~30 new test cases covering ordinals 1-1,000,000 and four-digit year forms
All 1,493 existing tests still pass
I’m happy to:

(a) submit it as a separate PR after this one merges, or
(b) push it to your branch as additional commits if that’s easier for review.
Either way, would love to see this Croatian PR move forward — it’s the closest thing the Slavic-language community has to a complete num2words implementation, and it’d unblock several downstream projects (TTS pipelines, accounting localisation).

Could a maintainer (@savoirfairelinux team) take a look?

[Jetman80]

(a) submit it as a separate PR after this one merges, or
(b) push it to your branch as additional commits if that’s easier for review.

@tihomirjauk I don't believe (a) can happen in the nearest future, it's been almost a year since I posted my PR. I just reference github source in my pip installs for now.

I am happy to merge commits if you provide PR to my fork, so this implementation and PR are more complete in this way. Thanks!