The circuit is designed to test battery chargers at my work. My company make's floor scrubbers, and these systems almost always have 2 12v deep cycle batteries in series, for a 24v system. The battery chargers on these machines is what needs to be tested. The largest chargers will put out 30A at 24V, and 25A at 36V. There is only one machine that uses 36V, but I wanted to have the ability to test this as well. However, since the vast majority of the chargers tested will be designed to charge 24V batteries, I wanted the most precision I could get, within safe limits. I am completely open to other ideas as to how to get there.
If you use the divider for the higher voltage system for both voltage systems, you will have a resolution of approximately 0.04V. Is this not good enough? That is 0.13% of full scale of the lower voltage. The theoretical maximum resolution would be 0.1% and circuit tolerances will make it impossible to achieve that anyway. The advantage is you eliminate the switch and a resistor and some program code to handle both cases.
Sound like a good idea? Or is my logic somehow off in left field.
That won't protect the Arduino if some accident or resistor divider failure occurs. If you want complete isolation and safety for the Arduino, you will have to isolate the voltage measurement as I suggested. It's a trade-off between the added cost and complexity of your circuit and the cost of an Arduino. It may not be worth worrying about, especially if you observe care in construction.