This PR fixes the bug identified in #1015, where specification of a list of signals as the input was not handled properly (each signal in the list was treated as a separate input rather than connecting a single input to the list). There was a similar bug in the output processing code. Unit test that exposes the bug included.

I think this code can be cleaned up a bit (to avoid the duplication in processing for the two cases), but want to get a fix out quickly since there is no easy workaround.

(Also found and fixed a typo in a docstring.)