This is a huge feedback! One point after the other:

• you can instantly share any python file with git, I don’t get the point, am I missing something?
• the python console does preserve the state. When you work with PyCharm you don’t have to recompute everything at all
• actually the pweave tool uses jupyter kernel behind the scene. All the magic will work
• Jupyter and venv: tried once, did not sleep for hours because it was hard to manage. On the other hand, starting a project with any interpreter (local, venv, remote Docker, remote SSH) is a piece of cake with PyCharm and once it is configured, you can even jsut click “install package” when you need one with the code checking tool
• bash integration? What about a plain subprocess ? But a proper python code should not use bash directly but API instead
• Why is it better to checkpoint without git? in any case git is a necessary tool for any serious data/dev scientist/engineer

In the end, I come from RStudio and switch to Python was hard in the beginning because I missed such a tool. I finally come to the conclusion that PyCharm with SciView is the equivalent for Python, sharing similar display and advanced IDE tools. RStudio is not only RMarkdown