Python similarly has 2-3 factions in my experience: teams doing engineering in Python and using all the modern tooling, linting, packaging, types, testing, etc; teams doing data science and using modern but different tooling (i.e. Anaconda); and teams that don't get onboard in any of the language health initiatives and are on unsupported language versions with no packaging, tooling, linting, etc.
Javascript/Node/Typescript has even more identifiable factions.
I think developing factions around these things is unfortunately normal as languages grow up and get used in different ways. Rust has arguably tried to stay away from this, but the flip side is a higher learning curve because it just doesn't let certain factions exist. Go is probably the best attempt to prevent factions and gain wide adoption, but even then the generics crowd forced the language to adopt them.
When you put it this way, personas might be a better term than factions.
Yeah I think that's a much friendlier term. I do think language ecosystems have a hard time, because on the one hand they should be aiming to be as useful as possible, which means doing more, on the other hand they have to acknowledge that any given user will likely not use all the language and that the rest of it may hinder them, which means doing less.
C++ does a lot, but has a big disengaged crowd, for many reasons, and that crowd will suffer from the push forward. Python and Node are similar.
The first two factions you describe in Python (types, testing etc. vs. data science and Anaconda) can work together just fine.
Source: I am in both factions, as are my colleagues :)