I was once working on a small tool that I tried different iterations of using a small variety of approaches; I tried implementing it in bash, perl, python, ES6 JS (classes and more), and a few other methods (I think at one point I even tried a LibreOffice spreadsheet).
But I kept getting mired and "stuck" in how I should architect things, or what framework to use, or this or that. I eventually decided that I wanted it to be a browser-based app of some sort (thus the road of ES6) - but then I started to wonder what tooling to use there. Lots of questions, etc - and I never got that far with it.
I eventually wised up and realized that it didn't matter what I chose - what I needed was the application to work. Who cares about frameworks, requirejs, css frameworks, etc - I just wanted something working. So I stepped back...and decided to chuck all of that away.
I ended up with a basic webpage, with a script tag and the code inside that - nothing fancy, purely procedural, old-school javascript functions, tied to events. Bare bones, basic.
...and I made real progress for the first time.
Basically, by regressing to the relative simplicity of Javascript back from the 1990s, I was able to get my tool finally going, without having to worry about the last two decades of cruft around that language. Now, I could have also done it using Python or anything else, in the same manner - but again, I wanted it to be browser based.
As you note, my own journey, and not something I'd tell anyone else to do (and certainly not something meant for a production environment!) - but for myself, and my need, this fit the bill. I'm still working on it; I figure when I complete it proper-like, I can go back and refactor it to something more respectable. Or maybe not.
After all, if it works, why try to break it? In the end, I want the results from the tool - who cares what it looks like under the hood, so long as it works.