nplurals=1 languages (ko/ja/id) lose count info on singular-vs-plural placeholder mismatch

[claude]

What happened

Translating ko (nplurals=1: "Plural-Forms: nplurals=1; plural=0;\n"), several plural entries where the English singular msgid has no placeholder but the plural msgid does (e.g. "There is one person ahead of you in the waiting list." / "There are %(count)d people ahead of you in the waiting list.") caused ./i18n fix to clear my translation.

_fix_format_errors's mismatch() check compares msgstr[0] only against msgid[0] (the singular form). For languages with nplurals=1 (ko, ja, id in this repo), msgstr[0] is the only string ever shown to users — at runtime it's used for count=1 and count=5 alike — so a translator naturally wants to reference %(count)d in it. But doing so trips the mismatch check because the singular English msgid doesn't have that placeholder.

The established convention in already-translated nplurals=2+ languages (checked de/fr/tr) is to hardcode "1"/"one" in msgstr[0] and only add the count placeholder in msgstr[1]. For nplurals=1 languages there is no msgstr[1], so following that same convention means the count is silently dropped from the UI for all values >1 (e.g. Korean would always render "한 사람이 있습니다" — "there is a person" — regardless of whether it's 1 or 50 people ahead in the waiting list).

Command that failed

./i18n fix ko
# cleared: 'There is one person ahead of you in the waiting list.'
# cleared: 'You have one more hour to borrow it.'
# ...and 7 more similarly-shaped entries

Impact

Not a hard blocker — I reworded these 9 entries to drop the count and match the singular's placeholder-free shape, validate/test now pass. But this is a real information loss for nplurals=1 languages (ko, ja, id) specifically on this class of string, and will recur for every future nplurals=1 batch that touches a "one X / %(count)d Xs" pair.

Suggested fix

Worth a design discussion rather than a quick patch:

• Either accept the tradeoff (document it in the instructions so future runs don't re-litigate it), or
• Special-case nplurals=1 catalogs in mismatch() to allow msgstr[0] to reference the plural msgid's placeholders too (union of singular+plural named vars), since that string is the only one ever rendered.

Flagging as a systemic pattern affecting ja and id as well, not just ko.