Fix Mongolian ordinal IndexError on zero

[patchwright]

Summary

num2words(0, lang='mn', to='ordinal') raises IndexError: string index out of range instead of returning an ordinal string.

Reproduction

from num2words import num2words

num2words(0, lang='mn', to='ordinal')     # IndexError
num2words(0, lang='mn', to='ordinal_num') # IndexError

Note to='ordinal' for other languages does not crash on zero, e.g. num2words(0, lang='en', to='ordinal') returns 'zeroth' — so this is an mn-specific gap.

Cause

In num2words/lang_MN.py, the ordinal suffix logic indexes the second-to-last character unconditionally:

elif number_str[-2] != '0':

For input 0, number_str is '0' (length 1), so [-2] is out of range.

Fix

Guard the index access:

-        elif number_str[-2] != '0':
+        elif len(number_str) > 1 and number_str[-2] != '0':

For zero, the condition short-circuits and the suffix falls through to the default 'дугаар' branch. That is the grammatically consistent outcome: 0 is not in the special set (1, 4, 9) that take distinct ordinal suffixes, so it receives the same default suffix as other non-special values (e.g. 10, 100).

Tests

Adds test_to_ordinal_zero and test_to_ordinal_num_zero, asserting num2words(0, lang='mn', to='ordinal') == 'тэг дугаар' and '0 дугаар' respectively. Both raise IndexError on the current code and pass with the fix.

[patchwright]

Gentle nudge on this one. The fix adds a length guard for deep negative indexing (x[-k] where k >= 2), which raises IndexError on short inputs. The test reproduces the crash on the current source and passes with the fix. Happy to adjust anything — just keeping it visible.