Faster processing in large datasets

[r-link]

Add functionality for faster plots for large datasets based on geom_bin2d() or similar concepts.

[r-link]

• seems to work sufficiently well with stat_bin2d() and stat_binhex()
• speed for large number of rows is several orders of magnitude higher than for pairs
  • example: 11 variables, 20 000 rows
    • ggcorrm() with stat_bin2d(geom = "point", bins = 10): 15.452 sec
    • ggcorrm() with geom_point(): 33.401 sec
    • pairs() with standard settings: 1443.283 sec
• further speedup while using ggplot2 probably only by precomputing the data in the plot in a faster way
  • easy for heatmaps
  • more complex for scatterplot matrices

[r-link]

• Maybe it makes sense to make a set of utri_/lotri_ functions based on stat_bin2d(geom = "point", bins = 10):
  • standard aes for stat_bin is mapped to fill,
  • mapping with geom = "point" leads to broken/duplicated scales unless fill is set to NA.