If the drive fails and you get a signal it’s perfectly valid to just let the default signal handler crash your process. Signals by definition are delivered non-locally, asynchronously, and there’s generally nothing to try/catch or recover. So handling this in Rust is no different than any other language because these kinds of failures never result in locally handleable errors.

That's not true - you can handle this pretty well with exceptions (yes it's nagging that you have to add them, but doable)... Not so much without.