Fixes #516

This ended up being trickier than I first thought b/c of different edge cases, but I think it's working well now. The core idea is pretty simple, though: We capture the x/ylab variables of an original plot layer and store them in our internal (tinyplot) environment. Then we match those up against any added layers that follow.

Examples

Errorbar (original layer) + ribbon (secondary layer)

pkgload::load_all("~/Documents/Projects/tinyplot")
#> ℹ Loading tinyplot
library(broom)
coefs = tidy(lm(mpg ~ wt + drat, mtcars), conf.int = TRUE)
plt(
  estimate ~ term,
  ymin = conf.low, ymax = conf.high,
  data = coefs,
  type = "errorbar",
  theme = "basic"
)
plt_add(type = 'ribbon')

Reverse case: Ribbon (original layer) + errorbar (secondary layer)

plt(
  estimate ~ term,
  ymin = conf.low, ymax = conf.high,
  data = coefs,
  type = "ribbon",
  theme = "basic"
)
plt_add(type = 'errorbar')

Flipped example: Error bar (original layer) + line (secondary layer)

plt(
  estimate ~ term,
  ymin = conf.low, ymax = conf.high,
  data = coefs,
  type = "errorbar",
  theme = "basic",
  flip = TRUE
)
plt_add(type = 'l')

^{Created on 2025-11-20 with reprex v2.1.1}