`ggplotly` uses colour that doesn't match the displayed gradient when providing one single numeric value to `fill` or `colour`

When converting from a ggplot object to a plotly interactive visualisation with ggplotly(), if there is only one numeric value for fill or colour, the wrong gradient colour is used according to the displayed legend. The legend uses a white-to-red gradient that doesn't match the blue gradient used in ggplot2.

library(ggplot2)
library(plotly)
# fill takes one numeric value
(single_value <- ggplot(mapping = aes(x = 1, y = 1, fill = 1)) +
  geom_col())

Convert to plotly:

# ggplotly(single_value)

Using two numeric values works well:

# fill takes 2 numeric values
(two_values <- ggplot(mapping = aes(x = 1:2, y = 1:2, fill = 1:2)) +
    geom_col())

Convert to plotly:

# ggplotly(two_values)

Using:

• R 4.1.0
• ggplot2 3.3.3
• plotly 4.9.3

Start by running the single-value and two-value examples in the issue and compare ggplot2 with ggplotly output, focusing on the fill or colour gradient and legend. The issue is done when a single numeric value produces a rendered gradient colour that matches its displayed legend, with a regression test covering both fill and colour.