Not sure you're still interested in an answer, but here are some thoughts.

'rules.value.list' seems to accept single values, or modalities, only. So your lines would not work. As I understand it, these rules are based on links between variables, conditional syntheses if you wish. This does not allow to restrict the range of values taken by variables, or give conditions and restrictions to single variables. I myself wish it would do so...

Maybe one of the reasons is that, by allowing so, values between the original and the synthesized datasets would differ so much that utility would be dramatically reduced (for instance if values like 900 or 1000 have to be synthesized to 700 maximum...). What you could is modify your variables in the original dataset so that all values above 700 actually are given the value 700?

As for your other issue, why not not synthesizing the variable 'net' at all, then sum the two 'payed' and 'received' variables in your final synthesized dataset? Sounds to me like the easiest solution.

Hope that helps