breaking(layout_column_wrap): Make width an optional argument

[gadenbuie]

Avoids items overflowing the parent layout container. With this rule, width becomes the point at which a new column is created, i.e. if width = "400px" then new columns are added to a row if there is at least 400px of space.

Breaking: This PR also makes width an optional argument and moves it to after the .... A code search shows that most people name the width argument, but in case they haven't we'll test to see if the first argument to layout_column_wrap() looks like a valid width value (a single numeric value or a valid CSS unit) and we'll emit a deprecation warning and use that first value as width.

Note: in the deprecation warning I'm assuming this breaking change will cause our next version to be v0.6.0 (in any case, we should review our deprecation warnings before the next release to make sure we have the right version in them).

Some code to try out this feature:

x <- lapply(1:3, \(x) card(lorem::ipsum(x)))

layout_column_wrap(!!!x)

# Explicit, named `width`
layout_column_wrap(width = 1/2, !!!x)
layout_column_wrap(width = "300px", !!!x)
layout_column_wrap(width = 400, !!!x)

# Implied as the first argument,
# leads to deprecation warning but gives same result
layout_column_wrap(1/2, !!!x)
layout_column_wrap("300px", !!!x)
layout_column_wrap(400, !!!x)

[cpsievert]

I'm on board with these changes, but I'd like to see this PR split into 2: one for the min(%s, 100%) change (with a clear examples/explanation of where it makes a difference, as well as a dedicated bullet for it in the NEWS) and then another PR for the breaking change.

Also, feel free to use your own examples, but here is one that's fairly simple, but demonstrates the change in behavior for the 1st change:

# Open me on a medium sized screen (i.e., between about 800-500px) 
# and notice how the cards "bleed outside" of their containing column

my_card <- card(lorem::ipsum(2), max_height = 100)

page_fixed(
  layout_column_wrap(
    width = "800px", 
    class = "border border-danger",
    !!!replicate(3, my_card, simplify = F)
  )
)

[gadenbuie]

@cpsievert I spun out the initial, non-breaking change into #851 and I'll rebase and re-word this PR after we merge that PR.

[gadenbuie]

Closing this PR in favor of a smaller, more targeted PR.