My point was that I find myself doing this:
tribble(
~rows, ~`col 1`, ~`col2 `, ~`col 3`,
"r1", 39.3, 45.5, 123.4,
"r2", 72.9, 87.1, 102.6,
"r3", 91.0, 88.5, 174.3
) %>% kable() %>% kable_styling()
and I don't particularly like to do it when this table purely exists as static text. If the answer is playing with the CSS, that's fine, I'm just making sure. I don't have a good grasp on where RMarkdown stops and pandoc begins. Or put another way, I don't know exactly what the limits of RMarkdown are and what can be accomplished with hooks. Therefore when I go beyond those limits, I'm always unsure I'm doing something in an acceptable way. For example, I looked into Pandoc tables and seeing how to assign a CSS class to them, and found this question/answer. Oddly it was there that I saw it mentioned that people will use JQuery to massage the tag attributes after rendering HTML, so that the CSS can be more specific. That's a little mind-blowing to me, because it's the kind of hack I do all the time, but that I don't like to do, because I constantly think there's a better way to do it and I just haven't seen the example yet.