I think that there are a number of things to consider, and some of these I am not sure exactly what the process is.
Firstly CC will only delete the data that the filename entry in the MFT points to at the time of deletion. I'm not sure of CC's secure deletion process, but possibly the data is overwritten, the file renamed to ZZZ.ZZZ, and the file then deleted. The point is that any other data for the filename that exists, for whatever reason, is not overwritten.
Secondly, when the file is overwritten by CC, does Windows create an 'undeleted' temp copy until the process is completed satisfactorily, and then delete it? Even if it didn't, the new (overwritten) data for the file is encoded by the disk software before being written to the disk, just as the original non-overwritten data was. This encoding expands the data by upwards of 5%. If the new data is larger than the old and won't fit into the old data sectors then it has to go somewhere else, leaving an old copy behind. Does Windows do a complete rewrite, to avoid fragmentation, or does it overwrite what it can of the old data and just stick a few new sectors elsewhere?
Thirdly, other common reasons for data to be duplicated somewhere on disk, and not have an entry in the MFT are edits, auto-saves, defrags, etc. Please add any more.
Browsing doesn't involve editing, but those jpgs have to be held somewhere. They could be in and out of ram, pagefile, hiberfile etc. This paging goes on, so I read, almost constantly, and I also read that the pagefile will be used to its capacity by Windows trying to be helpful no matter how much, or little, ram is used. Just watch those page faults mount when you're doing nothing.
It seems that this 'undeleted' data is found using deep scan. I've only run deep scan twice, and what it finds is surprising. Stuff I never knew I had and certainly never deleted. How it gets there is, it seems, rather a mystery.