Deja vu. I designed something like this 25 yrs ago. The problems I see are: 1. The circuit may behave differently if board positions are swapped. 2. You can't probe buried boards. Regarding #1, any sufficiently high speed board to board connections would have to be guaranteed by a custom deterministic path. Regarding #2, the active backplane could be replaced by a flexible version that would allow the boards to lie on their sides.