Erkki, I think you want the pilot's butt pretty much on the airfoil neutral point otherwise the pilot's weight overwhelms everything else when it comes to weight and balance. With a very light single-seater, you don't want the engine to balance the pilot, you want the engine to balance the plane like the paperclip on the nose of a paper airplane, and the variable weight of the pilot located where it will have little impact on balance. That's a big advantage of a tractor engine vs. a pusher for this approach.
Otherwise, I agree with you that an essentially constant-chord wing without twist using a straight center panel without dihedral or controls and outer panels with dihedral and elevons is the way to go. You may, however, find that you want to minimize the span of your ailerons to keep from washing out too much of the wing right when you need it most. As a rough idea, I'd suggest wing divided in thirds (left-center-right panels) with elevons taking up only half of the outer panels, so 1/6 span. You could then use vortex generators on the outer panels and/or stall strips on the center panel to tailor your stall behavior if needed.
For a 300 kg gross weight single-seat microlight using a proven flying wing airfoil like the Fauvel 14%, assuming a real world CL of about 1.25 you'd need about 13.5 sq m of wing area so 9 m span and 1.5 m chord at an AR of 6. With a 37 hp Polini Thor engine you should be able to manage a climb of over 4 m per second, top speed about 140 kph and a stall speed under 65 kph. All microlight-legal and very doable, keep the center section to 2 m span and you can have two removable and manageable 3.75 m outer panels for storage and transport.
PS--I like Sockmonkey's high-wing solution, which could get away with no dihedral at all, but I'd put the engine on the fuselage nose for easier access and probably go with tricycle gear rather than a short-coupled taildragger.