This PR fixes an obscure bug that got pointed out to me a year ago in e-mail from a colleague in Costa Rica:

I have a piece of code that I run for my students. I downloaded the latest control module and it does not print the root locus any more. I think it gets stuck in an iterative process. Nothing changed in the code. I suppose something in the updated version is causing the problem. Has anybody else made a complaint about the root locus?

Turns out that the problem was that he had a very unusual pole/zero pattern in which the algorithm for computing the default natural frequencies to use in plotting the grid was introduced in PR #138. The code essentially used the x-axis major ticks as the starting point for the natural frequencies and then added additional ticks to "complete" the frequencies on the y axis. The problem was that if the x-axis had tick marks that were orders of magnitude different that the y-axis, this could generate a lot of additional natural frequencies (which were subsequently removed a later point in the code).

I didn't want to spend a lot of time on this fix, so I basically put in a check to see when you might end up generating an excessive number of ticks based on this heuristic and switched things so that if that happens, you use the y-axis ticks as the starting point. There's a unit test to make sure it is working right (that also covers the new code).