Hi Manichean,
My main concern is performance. WIndows search (once indexed) is very fast when you have a colossal number of files. A non-indexed crawler in Python would be very very slow, to the point of unusability I suspect.
Were that not a concern, then yes indeed it could offer a lot of cool stuff, such as a button click to move all the files/folders to a working directory for importing and taking care of all the duplicate folder names etc.
I know I'm being very niche by only bothering to address Windows - but it is a niche plugin idea
. For instance unless it had some "direct" modes one of it's flaws is that you couldn't really use it to search for an author you haven't yet added to Calibre (you would have to create an empty book first).