It is ridiculous. Countless models later, and projects where you need to iterate on files is as unreliable as years ago. Yes, making the filenames unique by hand prior to uploading for each iteration solves it, it's also incredibly tedious. But I don't think the expectation that it should reliably differentiate between file names even with the same name is unreasonable. I uploaded a file to process, and it processed the file I uploaded a week ago.
If there are 2 versions, let us, and it see how is it differentiated so we can reliably reference it. Does the filename get a _01, _02, etc extension? Are they put in different folders? Why are uploaded files put into a black box in the first place? Let us see the file structure, let us edit them, let us manage them, let us organize them, let us do folders. Why is file management so incredibly basic?