I have now read this paper and I cannot agree that it refutes this hypothesis. Testing for fusion/segregation concurrent sounds does not test, if a person is able to detect ratios of frequencies.
One big issue with methodology of this study is that they assumed that liking/disliking an interval translates to consonance/dissonance. In some contexts (like horror film for example or certain music genres) I like dissonance more that consonance, since it better fits the theme. Liking something is influenced by a variety of factors. Another example: people generally like pretty things, but many people will dislike things, if they look too pretty. Culture can also influence what is likable.
It's an interesting study, but IMO, more research is needed before we can reach any conclusions.
The hypothesis that Tsimane people "are not able to hear them" is refuted in this paper: Perceptual fusion of musical notes by native Amazonians suggests universal representations of musical intervals https://www.nature.com/articles/s41467-020-16448-6