“Clear bug” doesn’t help unless someone can reproduce it. Possibly something else in your settings is contributing? Do you have any Advanced options set besides --snapshot-policy? Below is my result, running a backup twice with --snapshot-policy=required, then again with --snapshot-policy=off. It’s fine.
Can you come up with an exact setup of a new backup that does this? If so, please file an issue on it.
EDIT 1:
Creating a bug report is another option if you want someone to look at the DB timestamps for you, but actual file names are obscured for privacy. It might not matter much because you got about 30% extra opens, so sampling a bit might say whether or not there are some that have time issues. Because you can see the pathnames in your log, you can see if there’s a pattern, but first glance didn’t seem to find.
EDIT 2:
Lots of files seem to be considered ‘changed’ points to some tools to use if you look in the DB yourself.
Being on Windows, I assume this isn’t the mono issue (it started showing full time resolution) some got.