Add argument to nyquistplot for critical point

[Zinoex]

Considering a Nyquist plot for the return difference F = 1 + L, we should mark the origin as the critical point instead of -1. Similarly, it allows plotting of the determinant of the return difference for a MIMO system, as we necessarily need to test encirclements of the origin. Default remains -1 as the critical point.

[albheim]

Is it common to have the system 1+L instead of L, or is it nicer to plot for some reason? Couldn't you just plot F-1 instead if you know your system is on that form?

Not saying it isn't a valid request, just want to understand the need since I have not encountered this myself and I don't immediately see any real benefit compared to just sending in F-1.

[baggepinnen]

Maybe a good option would be to let the argument control the position of the point, so that

critical_point = -1

would be the default and

critical_point = 0

would mark the origin?

[codecov]

Codecov Report

Merging #688 (ccb0756) into master (355d1cb) will not change coverage.
The diff coverage is 50.00%.

@@           Coverage Diff           @@
##           master     #688   +/-   ##
=======================================
  Coverage   87.45%   87.45%           
=======================================
  Files          34       34           
  Lines        3579     3579           
=======================================
  Hits         3130     3130           
  Misses        449      449           

Impacted Files	Coverage Δ
src/plotting.jl	86.76% <50.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 355d1cb...ccb0756. Read the comment docs.

[Zinoex]

@albheim For a generalized Nyquist stability criterion, you plot det(I + L). Hence, you cannot modify the problem to encircle -1. Also, the Nyquist stability criterion is derived using the principle of the argument for encirclements around the origin for F = 1 + L, thus it may be desirable to plot F instead. I am by no means an expert in this area, but considering in particular the MIMO case, it is just a low-hanging fruit of 'nice-to-have'.

I have added the proposed change.

[baggepinnen]

Cool, thanks! The argument should probably be reflected in the docstring as well, and then we're good to merge

[baggepinnen]

Thanks for your contribution, a new release will be out soon :)