ggdist: Visualizations of distributions and uncertainty

Major changes:

• Require R > 4.0 due to several dependencies now requiring it.
• To complement the subguides for geom_slab(), there are now subscales, which can be used to adjust the limits or the expand parameter of the thickness per-geometry sub-scales (in much the same way those parameters can be set on scale_thickness_shared()) (#219).
• Default subscales and subguides can be set by assigning subscale_thickness, subguide_slab, subguide_dots, and subguide_spike in the global environment (#225).
• Allow use of square shapes in geom_blur_dots().
• More arguments to stat_slabinterval() now use waiver()s so that they can be set either in the call to the stat or in the call to the density estimator (#220).
• Bin edge openness can now be set via the right_closed and outermost_closed arguments to density_histogram() (#238).
• New helper functions interval_widths() and pretty_widths() make it easier to create nicely-spaced sets of nested intervals.

Documentation:

• Almost all parameter documentation (except some inherited from other packages) now includes an annotation listing valid data types for that parameter (#243).

Bug fixes:

• Stats and geoms now pass down check.aes and check.param to layer() so that aesthetics can be used to carry data to later stages of the pipeline without spurious warnings (#248).
• Ensure point_interval(<distribution>) supports tidy evaluation on the .point and .interval arguments (#221).
• point_interval(<tibble>) always returns a tibble() (#224).