I don’t think that’s the same.
That seems to be a setting for turning autocomplete on, ie. predicting what words you’re typing by what you’ve typed before, at least in English, don’t know about other languages.
The problem with the native Japanese keyboard is it is designed to type Japanese, including Kanji of course, and I don’t think it has any sort of mode that stops it suggesting which Kanji words you want to type.
Unfortunately, as far as I know, stopping an IME from suggesting kanji on kana input is not generally possible programmatically, in HTML or otherwise. Maybe on desktop (Windows, Mac, Linux, …) there are some sneaky ways to interact with the native IME somehow, but I’ve yet to see it done. If you want to stop all kanji suggestions, something like what you’ve done in this userscript is probably the only feasible way to go.
Essentially, the difference between auto-suggest/auto-complete and kana -> kanji suggestions is: who is providing the functionality? Auto-suggest/auto-complete is a service offered by the operating system, and most platforms will give you a method to enable/disable it for one input field. In HTML that’s the autocomplete attribute as mentioned above. On Android native, it’s setInputType() on EditText.
The kana -> kanji suggestions are offered by the IME or soft-keyboard implementation, not the operating system, and as far as I know, this particular feature is not exposed through HTML, Android native, and probably other systems as well. I am guessing our use case for this is so obscure that no platform has bothered to create an access path to this feature.