I think the Real Bug™ here comes from product-management: Nobody should be taking this kind of stochastic guess process and then just... 100% trusting the outcome with no feedback to the user and no way for the user to correct bad guesses.

For example, a prompt when opening the file like: "It's unclear what kind of data this is, here are a few options with a preview, pick which one you'd like me to use."

Annoying, but them's the breaks when you're making software and aren't willing to put in hard requirements about what it is expected to (not) operate on.

Ah, the times when computing used to be full of wonder and discovery.