ggpairs correlation plot shouldn't show p-values

[schmettow]

Using ggpairs on a set of metric variables produces correlation coefficients in the upper triangle. Since a recent update, this also prints p-values. P-values only make sense under the frequentist null hypothesis testing paradigm. I believe that, by default, exploratory plots should be paradigm-free. Especially, for everyone doing Bayesian Statistics, this command has become useless.

[schloerke]

Duplicate of https://github.com/ggobi/ggally/issues/377