There was a short period of time that I did exactly that, fetch the data from Nitter.
If I recall correctly, I was either getting rate limits from Nitter or asked to not put the additional load on their server. I also recall some downtime on the main Nitter instance which would take down the Twitter verification on Keyoxide.
Overall, we decided the Nitter intermediary was not worth it.
We could now try and host our own instance, relieving the burden on the main instance. The thing I don't like is that each Keyoxide instance may would have to selfhost their own Nitter instance as well then. Well, right now I am asking each Keyoxide instance to get a Twitter developer account and API token, but at least that's the "officially supported method to get Twitter data".