The diagram for these mutiple boilers is the most correct that I have seen.
The theory of pumping can be somewhat complex. Without going into great detail, I will try to explain some points here.
1. FYI, the glitch/fix article was written by John Siegenthaler, P.E. He has written alot about small hydronic systems and has several articals of this type.
2. Pumping towards or away from the boiler is not an issue in the particular diagram, due to the short primary loop. The added head to the boiler is minor. If your pressure is running that close to the relief valve setting, your expansion tank is probably undersized.
3. Heated water give up air more readily. Therefore the air separator is best placed after the boiler. On boilers with built in air separators, an automatic air vent on the air outlet of each boiler would eliminate the need of an inline air separator. However, since not all boilers have built in air separators, it is best to add one inline.
4. The expansion tank in relation to the pump suction is the most important details. Attached is an article from Bell and Gossett that explains why.
5. A minor benefit of pumping return water is the cooler water is less damaging to the pump and seals. Also, there is less chance of the heated water flashing into steam at the pump impeller (this is usually not a concern for smaller residential low head pumps).
5. For a point of reference, most packaged boilers have the pump mounted on the boiler pumping INTO the boiler. All packaged boilers that I have purchased are set up that way.
6. The mechanical codes I deal with require shutoff valves for each boiler. If the pump is dedicated to the boiler, it should not matter if the valve is before or after the pump. Most inspectors would not have a problem with this.