highlight() fails in lineplots after crosstalk filter if plot_ly()'s "color" argument matches highlight_key()'s key

I am making an interactive plotly plot with R using highlight() and crosstalk filter_select() functions. However, when the argument "color" of plotly's lines match highlight_key(), it becomes impossible to highlight any light of the filtered data in the plot.

Let's say you pick up the example in "Interactive web-based data visualization with R, plotly, and shiny" by Carson Sievert, in the Figure 16.10 of the chapter 16.1 "Graphical queries"
The example provided by the book works fine, you can highlight an item after filtering data:

library(gapminder)
g <- highlight_key(gapminder, ~country)
continent_filter <- filter_select("filter", "Select a country", g, ~continent)

p <- plot_ly(g) %>%
  group_by(country) %>%
  add_lines(x = ~year, y = ~lifeExp, color = ~continent) %>%
  layout(xaxis = list(title = "")) %>%
  highlight(selected = attrs_selected(showlegend = FALSE))

bscols(continent_filter, p, widths = 12)

However, when "color" argument of add_lines() matches highlight_key() variable (in this case, ~country) it is not possible to filter AND highlight the filtered data

library(gapminder)
g <- highlight_key(gapminder, ~country)
continent_filter <- filter_select("filter", "Select a country", g, ~continent)

p <- plot_ly(g) %>%
  group_by(country) %>%
  # Please mind that in the next line IS NOT "color = ~continent" as in the last example
  add_lines(x = ~year, y = ~lifeExp, color = ~country) %>%
  layout(xaxis = list(title = "")) %>%
  highlight(selected = attrs_selected(showlegend = FALSE))

bscols(continent_filter, p, widths = 12)

Is it possible to highlight items in an interactive plot_ly lineplot (because I've checked in boxplot and it works) when the color argument of plotly's plot matches the highlight_key?
I've also tried with geom_line() + ggplotly() but won't work neither

Commencez par la reproduction R minimale en utilisant plot_ly(), add_lines(), highlight_key() et filter_select(), en comparant color = ~country avec color = ~continent. Suivez la manière dont les graphiques en lignes gèrent le filtrage et la mise en évidence de crosstalk, puis vérifiez que les éléments filtrés peuvent toujours être mis en évidence lorsque color correspond à la variable de highlight_key().