Interesting! I always wondered what happened with really constrained 8-bit platforms like this when they tried to make them work for complex logographic languages with thousands of characters.
Also interesting that they chose katakana over hiragana, but perhaps the choice was because katakana is more readable at pixel resolutions? I don't know enough Japanese to know if it was based on a technicality of the language.
I don't think you mention it, but three of those characters are technically kanji -- the ones for day/month/year that are included.
Did you get a copy of the actual Jap ROM or did you have to retro-convert a Western ROM?
Regarding kana/kanji – well, my terminology may be fuzzy. But the emulator has "kanji" in the menu…
Regarding katakana vs hiragana: I guess, the simpler strokes were one of the reasons. Moreover, it seems that in daily use it can be distilled down to more or less the size of the Western alphabet. While there are 46 syllabograms in use, 8-bit computers managed to get away with about half of this. (The Sharp MZ-80 is another example.)
For fun, have a look at my own attempt at squeezing hiragana into an 8-bit character generator: https://www.masswerk.at/char8/#U3040 (A rendering demo can be found here: https://www.masswerk.at/rterm/ )