You can run "find duplicates" after each merge or entry. It's already automatically on when you make manual entries.
But,it is very imperfect and will not find all duplicates.
There are people known as John William at age ten and William John when adults. Known as Willie as a youth and Willis or William as an adult. Estimated ages from known facts for the the same person entered twice may be too far apart for the program to catch. Maiden names known from one merge, and not from another. Smith might not catch Smyth. Hallock won't catch Hollock.
etc, etc, etc.
If the program is made "looser", you will get too many false positives and spend all your time investigating "possible" dupes that aren't really dupes, that it simply is not worth it.