optgrowth.rst: PyPlot.jl converted to Plots.jl

[ajozefiak]

No description provided.

[arnavs]

Thanks @ajozefiak, this looks great. Few comments:

1. Going forward, it would be great if you could target this backend specifically (https://github.com/JuliaPlots/PlotsGR.jl), since that's what we'll be using. See below for this.

2. No need to try and match the LaTeX strings in the labels, titles, etc., since we'll be ripping those out.

3. There's some weird issue where our true and approximate value functions seem to be switched. I played around and I don't think this is a result of what you did (i.e., our approximate value function is genuinely below our true one). So I'll merge, but will post an issue.

[arnavs]

Actually, scratch that re GR. It is the backend we're using, but since it's baked into Plots.jl, there's no reason to do anything special for it.

And cc @Nosferican for the thing above. I'll confirm that the math is the same, in which case maybe it's due to something like shock stochasticity.

[ajozefiak]

Do you want me to completely disregard LaTeX strings? I can include them, the only issue was variable interpolation going from PyPlot to Plots.

[jlperla]

Yeah, I think so. We are going to have to go through a pass on all of the plots to get rid of the latex labels at some point in the near future, so if you strip out some of them it can't hurt. Sadly, unicode won't work well either in plot labels/titles (though it is great in the rest of the notes), so keep the labels as relatively simple text for now.

We can discuss the issues with unicode and latex issues in person sometime, but it is a fragile setup problem for many users.