I looked at outputs, not inputs, sorry. But, some comments are still valid.
Isolated, potential free, ground referenced etc.
I would ask them for a schematic of just the components connected to the real world for inputs and outputs. Usually, there is a compliance voltage that you can't exceed.
The reason why your dividers don't work out is probably because the input Z is not known and the compliance voltage is possibly larger than 10V. Usually to convert 0-20 mA to a voltage, you just put a resistor at the end of the line, so it may be much easier to use the 0-10 V inputs.
Figuring out the input Z of voltage inputs is easy. Put a high value pot in series with a 10 V source and adjust for 5V in. (1/2 Vin initial) Take pot out of circuit and measure the pot. That's the input Z.
So assume the voltage can't be more than 10 V at 20 mA or at 20 mA, the voltage read across the internal resistor is 10V.
I agree with Wade.
Simulate with a voltage source in parallel with a resistor, but this has input Z problems.
The pot trick works for voltage, but it might work for current only if you used a zero resistance ammeter.
Usually, I would expect the inputs to be 0-5V for 0-20 mA with a max allowed open circuit voltage of 24 V or so.
I looked at some of the docs and it's mostly a sell sheet and a sucky one at that.
In systems I put together, current output was used to set a voltage at the input of a device with the resistor at the device and the voltage was read at the device differentially. You really have to pay attention to ground loops.
The LM334 and The LT3092 are two possible simple parts to use. 4-20 mA simulators/calibrators also exist.
In your calcs you not taking into account the input Z. These are not ideal voltage sources (Z=0) and not ideal current sources (Z=infinity) and they are not zero resistance ammeters (feedback ammeters( with a voltage drop < 1 mV).
The current input may look like a 500 ohm resistor. e.g. 10/20e-3.
BTW: KISS works.