While I agree with your statement DxPL appears to have only a single focus at any one time, namely the directory/folder selected by the user. Past (missed) events are then discovered when the user selects a directory in which some action has taken place while DxPL wasn’t “watching” (that folder)!
I am almost 100% certain that is the way it works. I undertook ‘Rating’ and keyword changes from both machines on adjacent folders and nothing untoward happened.
Folder Monitor watches a number of folders at the same time, and their subfolders, based upon the configuration I input. The Python Folder Monitor I “developed” only watches one directory and there is sample code to make that multi-threading but I am learning to crawl before I walk before I run!!
So I believe that DxPL is keeping it simple, which makes sense but I have tripped over situations where an event has occurred, has almost certainly been trapped by DxPL (and by FM and PFM) but then ignored because DxPL applies a ‘Modified timestamp’ test and the software in question makes changes to JPGs without changing that field.
While other software traps the change DxPL ignores it and it will never be discovered.
I do not consider DxPLs simplicity to be an issue, I do consider not being able to detect what the majority of other software I have tested manage to detect successfully, is an error/ naïve coding.
PS:- I do find having to discover the inner workings of the "beast " by setting it test after test trying to fathom out what it is doing, to further my understanding of the workings a bit tedious, particularly when there is no-one from DxO to provide any insight. But it passes the time when the sky keeps dropping lots and lots of rain.