Not that I know of. Perhaps others might. Beancount doesn't touch your sources (which you probably knew), and I don't know of an external tool.
The closest I can think of is the dedupe functions in bean-extract. These should already recognize the transactions you manually entered, and comment out the duplicates found during the subsequent import, leaving you with just the the entries you didn't manually enter.
One idea is to simply use this workflow. Another is to extend the dedupe to feed into
autobean to modify the flags on your sources.