Add to_ordinal() and to_year() for Croatian

[tihomirjauk]

Follow-up to your Croatian language implementation in this fork (PR savoirfairelinux#643 upstream at savoirfairelinux/num2words).

Per your suggestion in savoirfairelinux#643 (comment) — submitting this against your fork so the implementation is more complete when/if upstream picks it up.

What this adds

• `to_ordinal()` — masculine nominative singular ordinals via last-cardinal-word substitution (prvi, drugi, treći, četvrti, peti, …, dvadeseti, stoti, tisućiti). Replaces the existing `raise NotImplementedError()`.
• `to_year()` — feminine-genitive last-word ending (the form Croatians actually use before "godine"). Collapses "jedna tisuća …" → "tisuću …" for years 1000-1999. Overrides the base class default that aliases to `to_cardinal`.
• Three lookup tables: `_ORDINAL_LAST_MASC`, `_YEAR_LAST_FEM_GEN`, `_ORDINAL_EXACT` (for exact 10^k cases like 1000 → "tisućiti").

Tests

• Replaces the `assertRaises(NotImplementedError)` in `test_to_ordinal` with 17 real assertions covering ones (regular and irregular: prvi/drugi/treći/četvrti), tens, twenties, hundreds, exact powers of 10, and full 4-digit ordinals.
• Adds `test_to_year` with 5 assertions for the genitive-feminine year form.
• All 1,493 existing tests still pass (verified with `pytest tests/ --ignore=tests/test_cli.py`).

Quick verification

```python
from num2words import num2words

num2words(1986, lang='hr', to='ordinal') # 'jedna tisuća devetsto osamdeset šesti'
num2words(1986, lang='hr', to='year') # 'tisuću devetsto osamdeset šeste'
num2words(2024, lang='hr', to='year') # 'dvije tisuće dvadeset četvrte'
num2words(17, lang='hr', to='ordinal') # 'sedamnaesti'
num2words(1000, lang='hr', to='ordinal') # 'tisućiti'
```

Linguistic note

Croatian ordinals decline by gender, case, and number. This patch only handles masculine nominative singular (the unmarked default form), enough for the most common "X-th" use cases. Full case coverage like Russian's `to_ordinal(case=…, gender=…, plural=…)` is a future enhancement.

`to_year()` returns the feminine genitive form because that's what matches the most common Croatian context: "X. godine" (in the year X). The last word changes ending (-i → -e for ordinals, e.g. "šesti" → "šeste"), and "jedna tisuća" collapses to "tisuću" for years 1000-1999.

Happy to revise per any feedback.

[Jetman80]

@tihomirjauk hvala!