On the fourth bucket, since I lived it: I shipped an app from a purchased template, it went through, and then everything I submitted for the next year was harder.
What I learned is that this is assessed per developer account, not per app. Once the account has a pattern of small similar submissions, new ones start from a worse position, and nothing in the rejection tells you that is what is happening. It reads exactly like your message - a guideline quoted back at you with no specifics.
If that is your situation, the things that actually helped:
- Ship less, ship bigger. One substantial app repairs the pattern faster than five small ones, and five small ones make it worse.
- Make the app visibly not-a-template. Custom UI rather than the stock components arranged the stock way, a real icon, screenshots that could not belong to another app.
- Give it something with actual work in it. Offline behaviour, a real data model, something that took engineering. The rule is aimed at apps that took an afternoon, and looking like one is the problem.
If none of that describes you and the app genuinely is substantial, the appeal route in the answer above is the right one and it does work. It is slower than resubmitting and it is the only thing that gets the decision looked at again rather than repeated.