You can pass Hot Coolant up and down too you know?
I just updated and tested it again. A reactor core will not pass hot coolant to another reactor core. I thought I could get a 2x2 working by arranging them in a snake pattern, with each passing both hot and cold coolant to the next one by putting both the primary and secondary facing pointing the same way, so I could feed cold coolant into one reactor core via pipe, have it snake its away around all 4 cores, then come out of the core next to the input. But only the last one emits its hot coolant to the pipe back to the heat exchanger, the rest just keep building up their internal buffer. So like I said, if you try to make place a 3x3 of reactor cores, there is no way to get the hot coolant out of the center one since it is surrounded on all sides by other reactor cores.
I got this 2x2 setup working by wrapping pipe around the side to collect the hot coolant. Notice that the 4 uranium fuel rods are centered in the inner corner of each of the 4 cores. With just the fuel, the output was 90 EU/t. I surrounded them with neutron reflectors and it went up to 181 EU/t, but was somehow unstable: it kept flickering between 0 and 181. When I replaced the reflectors with absorbers, I got a nice stable 225 EU/t.
So why idea why it isn't stable with the reflectors, and what is the difference between reflectors and absorbers intended to be? It seems like the absorbers are just better. Also I tried slapping another layer of cores on top and inserting neutron reflectors directly above the fuel rods, and the output didn't change. Shouldn't it have? I thought I tried the above/below reflectors last time and it did seem to help, but that may have been with all 4 fuel rods in a single reactor core.