Mask stage() at the expression level

[teunbrand]

This PR aims to fix #6104.
It is mutually exclusive with #6108 and #6107.

Briefly, this PR always masks the syntactic functions after_stat(), after_scale() and stage().
That way, you needn't prefix it with ggplot2:: if you don't have the namespace attached.

# Note: this PR is locally installed for this example
ggplot2::ggplot(ggplot2::mpg, ggplot2::aes(displ, hwy)) +
  ggplot2::geom_point(
    ggplot2::aes(fill = stage(drv, after_scale = scales::alpha(fill, 0.3))),
    shape = 21
  )

And users cannot override these functions, at least when used inside aes().

library(ggplot2)

stage <- function(...) stop("error")

ggplot(mpg, aes(displ, hwy)) +
  geom_point(
    aes(fill = stage(drv, after_scale = alpha(fill, 0.3))),
    shape = 21
  )

^{Created on 2024-09-17 with reprex v2.1.1}

[teunbrand]

Alright as a hail mary, I've changed from wrestling with data masks to changing the calls in the unevaluated aesthetics.
Thus, stage() now acts as any other function in that stage() doesn't work unless ggplot2 is loaded and ggplot2::stage() behaves as expected.
The only slightly insidious behaviour is that if another package, let's say {foobar} defines a stage() function, then this gets supplanted by ggplot2's functions, even if using foobar::stage() verbatim.

# Note: this PR is locally installed for this example
ggplot2::ggplot(ggplot2::mpg, ggplot2::aes(displ, hwy)) +
  ggplot2::geom_point(
    ggplot2::aes(fill = ggplot2::stage(drv, after_scale = scales::alpha(fill, 0.3))),
    shape = 21
  )

# We cannot use naked `stage()` when we haven't loaded ggplot2
ggplot2::ggplot(ggplot2::mpg, ggplot2::aes(displ, hwy)) +
  ggplot2::geom_point(
    ggplot2::aes(fill = stage(drv, after_scale = scales::alpha(fill, 0.3))),
    shape = 21
  )
#> Error in `ggplot2::geom_point()`:
#> ! Problem while computing aesthetics.
#> ℹ Error occurred in the 1st layer.
#> Caused by error in `stage()`:
#> ! could not find function "stage"

^{Created on 2024-09-23 with reprex v2.1.1}

[thomasp85]

LGTM. I don't suppose there is an elegant way to test this?

[teunbrand]

The masking mechanic is testable, using ggplot2::stage() is testable, but the 'if ggplot2 is not loaded' condition is hard to test.

[thomasp85]

Well, that is probably sufficient :-)

So please add a test for the namespacing