Hello Teddy, there are examples for multiple methods for a benchmark system with two molecules, which of course you can easily adapt to the two atoms NaCl system:https://github.com/Colvars/examples
Umbrella sampling and in general methods based on biased probabilities require WHAM or another unbiasing methods. Axel mentioned already the widely used implementation by Grossfield, there is also one in the GROMACS analysis tools.
Can you elaborate on what is not working with ABF? For such a simple system, I find that unlikely unless there are problems in the setup. In that case, another sampling method is unlikely to give better results.