Read literally, the word "clean" in a coding context is most naturally interpreted as "observing a consistent style at a basic syntactic level" (e.g. camel case vs snake case), or "having a minimum of irrelevant clutter" (e.g. variables and functions that are defined but never used). It's about the sort of things that a linter looks for.

Unfortunately, "clean" has been stretched and concept-creeped to the point of meaninglessness in software engineering, especially by the book of the same name. Now, "clean" has become an invitation to enforce ideology on complex questions we don't understand. Once something is declared not "clean", action must be taken to mitigate the primal fear of contamination, even though we don't really know what "clean" and "dirty" mean in a coding context.

We should stop using this vague crutch word for everything we approve and disapprove of in code, and find more accurate words to express our feelings. If we think code is repetitive, we can call it repetitive, and have a conversation about whether repetition is good or bad in this context.