So here is what I was thinking.
Sump with refugium - no filter floss / fine media prior to the fuge section
4-5 inches of fine grain sand (in fuge)
Flow across the top of the water column only - allowing detritus to settle
floating plants to remove excess nutrient from water column
Java moss to supply supplemental diet for the fauna
eventually crypts to lower base nutrients in the substrate
Red cherry shrimp or pond snails to eat the dying floaters / java
black worms or tubifex for substrate aeration / breakdown of settling organics.
In this experiment I would like to have a dominant population of red cherry shrimp as I've read that older RCS will eat black worms - however blackworms will most likely out-reproduce RCS if given the chance. It will take a while before detritus starts seeping into the lower levels of sand, so I think a few months with just RCS would do it. Tubifex worms may be a better option as they will stay out of the water column and may burrow deeper into the substrate.
Before you go whole hog you might want to do a proof of concept with a small tank.