Unicode now handles all the Kanji in JIS. I wouldn't be surprised if Ruby predated that. It almost certainly predates good library support for all the Kanji in JIS.
I think the problem isn't whether it handles all the Kanji in JIS – it does. But the problem is that JIS at the time was so common that it didn't necessarily make sense to settle exclusively for then-less-used UTF-8. That would make re-encodings necessary at interfaces and on IO.