The radiator can take 4 120 mm fans. [Edit: was 150 mm ?] There is a smaller model [edit: with room for 3 fans], but not much cheaper. It's so big, we can probably get away with no fans at all. We have space for a wide radiator in front of the air conditioning radiator, so that's no problem.
We chose the combined reservoir and pump arrangement, which makes for one less thing to mount, and one less connection to make. We had to remove the top from the pump, and replace it with a different top (and uses different screws, provided), and that top adapts to the reservoir. We decided that we had to test it out with clear water. Alas, a 12 V @ 1A plug pack didn't seem to want to start it. When we looked up the specs, the pump needs 20 W, no wonder it didn't start. However, a larger plug pack (65 W) also failed to start it. This is getting silly. We checked the web for connections; it seems all we needed was ground on the black wire and +12 V on the yellow wire. The blue wire seems to be for the tacho feedback, even though on PC fans, blue is supposed to mean PWM input.
Ok, we have a few lithium cells with some energy in them; Weber connected 4 of them and at the same time tested the spacing of the new BMU boards. The first shock was that the links were about 5% too long. It appears that all the links that came with the original shipment of cells are like that. We'll likely have to re-kink them or drill new holes. Sigh. Fortunately, the links for the last 16 cells (which we got from EV Works) are the right size. We figured that 4 cells would provide 13.2 V, which is at the upper end of the voltage that the pump should work on.
No pumpa da wata.
In desperation, we pulled the pump apart. Sure enough, the PCB tracks to black and yellow are thick; the track to blue is thin. Surely black would not be +12 V. We decided to risk running the motor with no load off half the pack. No go. Well, that's below the minimum voltage; we tried 3 cells. Still no go.
But wait! With three cells, after a 4 second delay, the rotor did spin! Yay! Back to 4 cells: no spin. When we put the pump back together, and put in some water, it pumped on 3 cells but not on 4.
When I got the pump home, I put it on my small power supply, and found that it would start on voltages below 12.98 (I didn't check how low it would go), and it would definitely not start on 13.0 V. I guess with PC mother boards, 13.0 is considered out of spec. Yet the pump specifications state that it will run on up to 13.2 V. In fact, it happily runs to over 14 V once started, but won't start above 12.98 V (according to my cheap Dick Smith meter).
Since this is awkward for running in a vehicle, I set the power supply to 13.8 V (the highest we're likely to see) and tried starting with one diode from a power bridge. No go. But with two diodes, it started every time. The quiescent current must be enough to cause a little over 0.8 V to be dropped across the two diodes, which is enough to allow it to start. Sheesh. But that's reasonably practical. We'll have to keep an eye on it though; the last thing we need is for the cooling to not start and our controller to overheat. Well, I'm sure it won't die, just cripple itself with low output.
Edit: added photo of cells and boards.
All those packages. 
With the PCB oriented such that the micro is at the top, the one you need to change is just to the left of the micro. It's marked 4701 on mine. 