I'm really not sold on the delta configuration (although I don't actually see an issue with the folding wing proposal) as it isn't the best for stable gliding or control. Why not try for something more akin to the Barnes Wallis Swallow swing wing? Suitably modified, of course.
That way you can get a nice, long lean rocket phase and a stable, broad spanned traditional glide phase while keeping the weight down nicely. Cut the wing sections from foam and skin them, as per current RC aircraft, and add a carbon fibre stiffening rod to each which will also give you something to attach the pivot to as well as a lever for the opening mechanism which could then be as simple as a restrained spring hooked up to a yoke for simultaneous opening. The catch releases, wings go *SPROING* and hey presto! one glider.
Sticking with the tractor rocket motor proposal, rather than have a separate stage that is detachable mount the rockets to a Y shaped canard on the nose of the aircraft. The lower fin shouldn't create too much drag, the upper two should provide a touch of stability owing to the dihedral effect. Mount the rockets in a faired pod at the end of each canard and they should be far enough out from the fuselage that a simple foil sheath will protect what's underneath.
A butterfly tail at the rear should provide the rest of the dihedral stability and at least partly counter a flat wing as well as providing the rear control surface without having to go down the elevon route.
If you use a lifting body fuselage then you should get greater lift, better stability and greater load capacity than a simple cylinder. The mass of the fuselage can be kept low by shaping it from expanded polystyrene using a hot wire cutter, solar film skin and/or foil to keep a smooth surface as well as it's insulating properties preventing low temperature damage to the electronics gear inside it. Add carbon fibre stiffeners where required to take the structural load without increasing the mass too much.
6 Aerotech E28s should give a nice amount of oomph for the boost phase as the model wieght ideally would be below 3 KG, even with all the electronics gear (approx 235 N thrust for 1.2 seconds). Alternatively use long burn ones like the Apogee E6 with a 5.8 second duration and a similar amount of overall thrust. You could even combine the two for an initially larger boost to start with going to a gentler, sustained boost afterwards.
I believe that the abopve would be a touch more efficient but also should look significantly more elegant - think along the lines of a B1 fuselage with a Y canard added to the front and a butterfly tail to the rear rather than a conventional tailplane.