This is true of broadcast comps as well and has been since the 90's. Essentially dynamic processing that uses slow RMS detection rather than peaks is going to respond this way.
I think there is a lot of inertia with loudness unfortunately.
On an aside, I don't care for the DR system because it's based on some non-adjustable measure of crest factor which is naturally going to be less forgiving of music with steady content like high gain rhythm guitars.