It's not more popular because Dan Luu actually measured this [1] and those languages aren't better in practice.

A priori, it does seem strange. But the relative Haskell vs Python tokens in the internet go a long way to explaining it, as another commenter has said.

[1] https://danluu.com/pl-tokens/

My main takeaway from that post was that, apart from "don't use assembly", none of the experiments showed strong evidence either way, and the things that took time were when the agent got stuck doing silly things like invoking cargo with the wrong arguments. Ie. there's a lot of measurement noise, and not any strong signal, and "LLMs are good at generating $PL" isn't well-founded as an argument for picking $PL. So we're back to the old arguments for picking a language, like "I find it readable/enjoyable"