DAM is more re-active than non-DIT cars intentionally - "premium" fuel is not required for DIT models. So what may have been some FLKC in the non-DIT cars, might result in a drop in the DAM in the DIT cars, for example. Just like you can see some random FLKC (possibly due to false knock), you should also not be surprised if there's a random drop in DAM in the DIT cars. How long it takes to come back up in not indicative of how serious the original knock events and does not mean the car is still knocking. In order to advance the DAM, you have to drive in specific load/RPM ranges without knock. So, what is important is when the DAM is dropping (and how low) and what other knock responses look like overall (FLKC and FBKC). If a car has poor quality fuel, in most cases you are going to see a lot of knock activity on a WOT pull.
Subaru knock detection is like using a large net to catch fish. You'll definitely catch what you want to catch (actual knock) but you will also catch what you don't want (false knock). If you used a small net, you would get less false knock but might actually miss actual knock. So, it is a huge compromise based on the cost of system. You have to look at the entire picture when gauging if there is actually an issue.