Thanks for the detailed post, although I think it's too strong of a statement to say "do not use" shadow files. They are a great solution for many people, but the have their drawbacks - along with every other workflow.
The biggest drawback is that users have no control over how long it will take PD to finish generating them, and many don't wait (or know that they should wait) for the yellow icons on the media library clips to turn green. Once that's done however, the user never has to take any additional steps to produce the full quality video.
The biggest drawback of using MagicYUV or another intermediate codec is that the file sizes are huge, because the converted AVI files are uncompressed. In your example of (4) 50GB 4K clips, the MYUV versions might need more than 1TB of SSD space, and that makes them impractical for some people. If space is available, though, PD easily works with those streams in full resolution and there are no additional steps to produce a final video.
Your approach has many more steps both before and after editing to get the full quality video. That's because you're duplicating how PD makes shadow files, and you have to do each step manually instead of allowing PD do all the work. I'm sure it's quicker to get started than waiting for shadow files, but you're also stuck with the problem of working with low resolution clips the entire time.
For many projects, that isn't much of an issue, but if you're making critical edits, like masking and overlaying clips where edges have to be perfectly aligned, you may not be able to achieve the right alignment
at all because moving a 640x360 clip just 1 pixel is the same as moving a 4K image 6-7 pixels. There's no in-between, and no way to move the final image by 3 pixels, for example.
Like I said, each of these three workflows have pros and cons, and there really aren't any right or wrong choices. I encourage PD users to look at each project upfront and think about the kind of edits they'll be making before choosing one of these helpful options.