It could be a big burden on the CPU as if you add a file, and you have thousands of peers they could all ask you for the same data over and over again, at the same time limiting the troughput. Having to cache what you checked and what you haven’t is just as racey, as well as takes up ram.
The alternative already exists, and it’s touching the files. You can touch them, rescan and set the dates back to where they were.
In the current situation those thousands of peers (how many people have thousands of peers?) keep retrying to sync the files over and over generating a lot of traffic. But I get the point that you are not interested in my opinion or investigating solutions, so I’ll leave this discussion at this.