Focus all your effort on catching errors that a linter and auto-formatter could catch. Shrug when questioned about architectural decisions, what really matters is that this field was PascalCase when it should be camelCase.
When questioned why this isn't caught automatically by the build process, you have a few stock responses (Bonus: These may well be true!):
- Linters are too slow, and a fast development experience is crucial
- Linting the entire existing code base would effectively destroy "git blame" as a useful tool.
- Different teams can't agree on whether to column-align wrapped function arguments, so they've agreed to disagree on this.
So instead of automating away the mental overhead of checking for spaces vs tabs, or LF vs CRLF make sure you go over the PR with your own linter and nitpick every violation, even the suggestion level ones before you even consider trying to work out the validity of the submission.