Some "numbers" having readings removed aren't actually numbers

[ahlec]

We have the config option to remove readings from numbers, which we're currently doing by removing readings associated with 一二三四...

However, not all occurrences of those characters are outright numbers. Example: 一通り (ひととおり) uses 一, but shouldn't have its reading removed because it's part of a phrase.

A potential first step could be that we only remove readings from numbers where the character is 一 and the reading is いち, いっ, etc. But that might not work long-term. In order to fully fix this, we might need a separate tool that has a database/dictionary lookup to determine if a word is a number/number + counter (we want to remove the reading), or if it's a regular word (we want to keep the reading).

EDIT: Interestingly, 一先ず doesn't remove the reading from the 一. So clearly this problem isn't 100% universal even currently.

Actual: 一通りの一先ず → 一通[とおり]の一先[ひとま]ず
Expected: 一通りの一先ず → 一通[ひととおり]の一先[ひとま]ず

[ahlec]

Another test case to include in this would be 一切, which would also challenge the earlier proposal of "only remove it if the reading is a regular one (here, いっ)."