Fix labeller to pass .default for lookups (#4031)

[waltersom]

Pass .default to as_labeller even when margin_labeller is null, such as
when using named arguments for a lookup table.

This should make the behaviour match the documentation.

Fixes #4031

[karawoo]

Thanks @waltersom! Could you please put space around the = sign (following tidyverse style), and add a bullet to NEWS.md that briefly describes the change and ends with (@​yourname, #issuenumber)?

[clauswilke]

While we're touching this code, I think we should also consider whether .multi_line should be forwarded to as_labeller as well.

[waltersom]

@karawoo I have made the requested changes.

Also @clauswilke, I was looking at also passing .multi_line, however this was more complicated, and more work is required for that. A problem is that if .multi_line==FALSE, when using label_parsed, collapse_label_lines is eventually called on the output, which converts any expression()s to characters. Further, label_parsed adds list() around the argument to expression(), which I don't fully understand.

[karawoo]

Hmm, I see what you mean @waltersom. The .multi_line issue is more complicated. Using the example from the original issue with current ggplot2, .multi_line = FALSE gives the plot below, and this PR doesn't change that:

library("ggplot2")

tdf <- data.frame(
  x = c(1, 2, 3, 4),
  y = c(4, 3, 2, 1),
  a = c(0, 1, 0, 1),
  b = c(1, 2, 2, 1)
)

ggplot(tdf, aes(x, y)) +
  geom_point() +
  facet_grid(
    ~ a + b,
    labeller = labeller(
      a = c(`0` = "a^2", `1` = "a[2]"),
      b = c(`1` = "One", `2` = "Two"),
      .default = label_parsed,
      .multi_line = FALSE
    )
  )

If we just passed along .multi_line to the line you've edited, it becomes:

ggplot(tdf, aes(x, y)) +
  geom_point() +
  facet_grid(
    ~ a + b,
    labeller = labeller(
      a = c(`0` = "a^2", `1` = "a[2]"),
      b = c(`1` = "One", `2` = "Two"),
      .default = label_parsed,
      .multi_line = FALSE
    )
  )

I'd be inclined to merge this and open a separate issue for the .multi_line problem, but what do you think @clauswilke?

[clauswilke]

This seems fine by me. The only reservation I have is that I don't know where we are in the current release cycle. Is it fine to merge into master or should we wait?

[karawoo]

Yeah, let's wait until this release is finished

[karawoo]

(CI failures are unrelated)

[yutannihilation]

(For failures, macOS runner will keep failing for a while until the R 4.0.2 installer is reliably available.)