updated uniannova and ttest with leveneTest, aligned digits of output…

[LKobilke]

This PR comes with several updates:

1. new data:

   • incvlcomments.rds + documentation
   • snscomments.rds + documentation

2. new functions with features:

   • tab_percentiles() + visualize(), closing #30 [I opted for a quantile plot]

3. updates for existing functions:

   • updated test_icr() works with grouped data, also covered in testthat(), closing issue #17
   • updated t_test() and unianova() to run a leveneTest and set var.equal = FALSE depending on test result [also included a var.equal-parameter deprecation warning]
   • updated unianova() with var.qual = FALSE to use test.welch() and games-howell posthoc
   • updated the output of the posthoc tibble to show the user more telling column names
   • updated t_test() and unianova() to display tibbles with 3 digits, closing issue #34 [I opted for three digits because it didn't significantly affect the length of the tibble and proved to be more versatile. This is especially true since the three-digit format overwrites the default seven-digit scientific notation that tibbles use outside the console]

4. updated warning messages:

   • icr.test(), t_test() and unianova() now raise special warnings when no variables are provided by the user

Note: The devtools:check() function still generates two notes. The first refers to the dependency on imported packages, which makes tidycommmore susceptible to deprecation errors. The second note is about the use of, for example, vars instead of .data$vars. However, changing to the latter contradicts the new evaluation procedure in tidyselect 1.2.0. Consequently, I have opted for the former approach.

[LKobilke]

Mental note:

• functions for categorical variables are collected in "R/categorical" (tab_frequencies(), crosstab()), but functions for continouus variables are not organized in a similar manner, e.g.: describe(), tab_percentiles()
• both describe() and describe_cat() are part of "R/describe"