I know what a half-normal patchbay scheme is; I've wired plenty of "realworld" system patchbays myself. And the behavior you described above is what is emulated in the example I gave.
It may be confusing you that I have the put the inputs at the top and the outputs at the bottom, but that is because the abstraction emulates a patchbay-wired -device-, with inputs at the bottom row of one patchbay row pair, and outputs at the top row of another. It would be simple to separate the input and output mechanisms from my example if it's more intuitive or practical for your purposes.
It also may be confusing that I've given a single patchbay output "jack" two outlets and a patchbay input "jack" two inlets, but if you think about it, something like this is necessary, because you have to deal with the "normal" signal flow that happens within the bay as well as any patch cord connections you make. If you wish, you can hide these normals behind the scenes with [send~] and [receive~] objects so that they don't have visible inlets and outlets to more closely emulate the physical appearance of a patch bay (normals hidden), and this is possible, but a little more work than I was willing to do for an example this morning.