The best answer can change over time (and should be re-assignable as such); and the asking person may be the best resource to eventually answer their own question (it is a problem dearer to them than to us).
That said, would it be considered an appropriate action to automatically flag for review a newly minted self-accepted answer in the event that the question was answered three or more months prior?
Flagging for review doesn't imply that the user did something wrong, it just implies that a moderator would need to inspect the situation to determine that the user either:
- Found some new solution, detail, or evidence that improved upon the answer and as such should be captured through the existing self-answer mechanism.
- Determined that accepting their own answer with excerpts from a previously existing answers could benefit them in some way. (I already know that it won't give the acceptance points).
While the latter scenario borders anti-social behavior, a human needs to weigh in on whether anything truly odd occurred; hence the idea of auto-flagging instead of taking some (yet undetermined) action.