`tableby()` support for `stats::ttest()`

[jmbarbone]

I know the docs for tableby() suggest that the anova is equivalent to the t-test, I think it's important to still have the option. We already have access to the stats::wilcox.test(), which is only valid for 2 groups.

Happy to do the work and submit a PR.

[rotegrutze]

Hi jmbarbone, I am also just learning this package. If you are looking at one-sample paired t-test (which most of my work in cohort evaluations), try the paired() which will properly build a pre-post treatment table with difference and p-value (paired t.test)