AFAIK the "right" way to do this is use a virtual device for system audio, and have that in turn output to the actual audio interface, so you can keep it maxed out.
But that doesn't really feel like the ideal solution.
That said I'm probably missing something. EasyEffects is able to process audio in ways that should enable what you're asking for, I just don't know how.
Also, if you haven't already, check out qpwgraph. It's a virtual patch bay for routing audio around however you want.
Best part, is it's not an add-on, really. It's just a GUI for capabilities that pipewire already has.