Yes, this is a review of just the base language, not any additional tooling. As for particular points:
According to the surveys, overwhelming majority of developers of most languages use language-independent tooling like VSCode, Jupyter, Vim, Notepad++ etc., so I think it's reasonable to not review language based IDEs.
Similar trend seems to be happening with data scientists, and RStudio lost more than half its use share from 2018 to 2021 according to Kaggle surveys.
It would be more accurate for me to run these in a Jupyter environment than with VSCode and terminal like I did, I didn't do this as it would make formatting it for the post more difficult.
The working theory for this series is that language that struggles a lot given very simple tasks will likely struggle even worse given more realistic tasks. This isn't really provable, and it might not work for some highly specialized languages (like let's say Verilog or WebGL Shader Language), but I think it's very reasonable for data science.
According to yet another survey data preparation, which is basically a very large number of such simple tasks, is the most time consuming part of the job, so I think it's fair to see how well the language is doing on such tasks.
And in the end, I rate languages relative to others in similar domain. So I'm not saying it's impossible to do a decent job with R, just that other languages (Julia and Python in this case) seem to be a lot better at the same thing.