This issue is to keep track of some items triggered in PR #522 and described in issue #543 related to computing the frequency response for a discrete time system that has poles or zeros on the unit circle. For example, if a discrete time system has a pole at -1 then Bode plots still works but there is a gap in the response near the Nyquist plot (from the nan's). This also generates a warning message, which is what triggered this particular issue.

We should figure out how to "properly" handle plotting frequency responses for systems that might have an actual singularity at one of the points of evaluation.