Per Splatz's original post, the problem was that the inputs were not isolated. Looping the current signal was not an option. OK, leaving one set for current and the other set for voltage might confuse the next guy if he doesn't truly understand analog signals. That said, set them both to voltage and use a common shunt resistor. 250 ohm for 0-5v(1-5v actual) or 500 ohm for 0-10v(2-10v). The inputs when set to voltage will be some very high impedance so paralleling won't be a load issue as they are still significantly higher then the shunt resistor. The transmitter will never know the difference. It will still be controlling current. Per Splatz's comment, the inputs are on the same controller so they are probably not very far apart, so you will only have a voltage signal essentially between them. Hummmm, single, 25 cent component or $75 or more loop splitter? Since this is storm water control, my bet is NASA grade calibration is probably not necessary.
A surge arrestor anytime your transducer is outside is never a bad idea.
Ive done this at home on arduino's but im not sure i would ever do it in the field unless its a patch to get me through the night.
Honestly if a customer is using a engineering firm who sub's the job to a contractor $75 wont cover lunch. Its a dam good idea looking for a problem.