Ok, after some digging. This looks like a known issue:
https://github.com/JuliaData/CSV.jl/issues/912
It comes down to the situation described here:
https://github.com/JuliaLang/julia/issues/35663
I have a few recommendations to get around this. Basically, you should make sure you activate your working environment as the very first thing you do in the script, before any using or import package statements. Also, update your packages in your base environment (v1.6) to the latest possible versions.
Finally, if you are using the graph_env environment, make sure your IJulia kernel in your workbook is using it as the default:
IJulia.installkernel("Julia GraphEnv", "--project=GraphEnvProjectPath")
where GraphEnvProjectPath is the path to your project folder, and select this kernel for your workbook.
Alternatively, launch your Jupyter notebook session in the REPL from that project environment folder (Using IJulia · IJulia);
julia --project=~/.julia/environments/graph_env
Also, are you using VirtualEnv.jl somewhere? I’m not familiar with this package but it might mess with some of this native environment management.