Improvements to plotFit()

[mariusbarth]

Hi @danheck,

This PR adds support for for posterior predictive standard deviation (stat = "sd") and correlation (stat = "cor") plots in plotFit().

I implemented these additional plots because the covariance plots can quickly become unwieldy, and I observed, in some use cases, that these plots were dominated by variances (making co-variances hard to discriminate from zero).

(I think) I also fixed the placement of separator lines for covariance plots. Here is the new behavior, where the lines separate each column of the lower triangle of the covariance matrix.

I could not make sense of the old behavior here:

There is also a tiny change in PPP.R. This change is unrelated, but harmless. It does the same as the old function, but is faster (magnitude 2x). If you want to go one step further, you could use asplit(), but this comes with a dependency (R>=4.1.0).

If you have any questions or suggestions, feel free to ask.

Best,
Marius

[danheck]

@mariusbarth That's great, thanks a lot!

A plot of estimated correlations is certainly useful.