I have found that water moves through the loop fast enough that it's heated only a fraction of a degree in each block. In effect, the whole supply of water shares the heat carrying load. It has to do with the particularly high specific heat of water. The net power dumped into the water compared to the net power radiated determines the water temperature rise - rise of the whole water supply. The advantages of ordering the waterblocks according to importance - CPU, GPU, RAM, PCH, VRM etc. - are finite but small.
Splitting parts of the loop into parallel branches works sometimes. Each branch gets only part of the flow. If the waterblock works well with lower flow, splitting is okay. Most CPU and GPU blocks work best with higher flow, which argues for a single path through those. On my testbench, I plan to split RAMs into parallel paths because RAM doesn't dump much power and can stand the lower flow in each branch. The set of RAM branches is still downstream from the CPU and GPU.
Tubing sizes are spec'd and spoken of by their inside diameter, whether inch or mm. I don't like compression fittings either. I use Koolance barbs to match the tubing ID that can swivel to avoid tubing twist unscrewing the barb. Each Koolance barb comes with a clamp. Note that the clamps have to contend with the tube's outside diameter. The standard is 3/8 ID and 1/2 OD, but some 3/8 tubing has 5/8 OD. The ID has to match the barb and the OD has to match the clamp. This page has barbs and compression fittings:
http://koolance.com/index.php?route=product/category&path=62_63_99_161Quick disconnects work well to let you remove a component or add a drain tube. They shut off water flow with only a drop or two leakage.
I use distilled water from the grocery store - about US$ 1 per gallon (3.7liters). I add copper sulfate solution, about 1 drop per liter. That's marketed under the name "Dead-Water" at FrozenCPU.
Slowing the fans when the water is cool cuts down the noise. That's the sort of personal preference we each decide for ourselves.
I have found that pump pressure is spec'd in whatever units the writer felt like that day. Conversion is a pain, but necessary for comparison or calculating expected flow.
1 psi = 27.68 in of water = 0.703 meters of water.
See the Koolance page for their version of the D5. They call it PMP-450. (Mine still have the Laing label plus Koolance's rebranding.) Their pressure is in meters of water. The real fun happens when pump and block pressures are mixed between psi, inH2O, mH2O and flow rates are mixed between liters per hour, liters per minute, gallons per minute etc. Excel spreadsheets are my friend.
Jeff