While working on the half-width support, I noticed that IsKatakana (as well as IsHiragana) are based on golang utf8 tables, and the range being used a probably too wide for what a japanese speaker would consider being a katakana or not.

For example, this was slightly unexpected:

• IsKatakana("ㇰ") = true // attention ㇰ != ク and ㇰ != ｸ
• IsKatakana("ｳｶ") = true

This even more:

• IsKatakana("㋾") = true

And I would say this is wrong:

• IsKatakana("㍓") = true

IsHiragana as fewer kirks, but still funny yet unexpected behavior. For example:

• IsHiragana("🈀") = true
• IsHiragana("𛁟") = true // \u1b05f

For IsKanji, I also suspect the range to be wider than what would make sense to a human reader, but considering the difficulty to put any kind of boundary to it, I've skipped checking.