ggplotly() creates a wrong legend in some cases

Given

library(ggplot2)
library(plotly)
d1 <- data.frame(X=1:10, Y= (1:10)^2, b=paste0("B",1:10),d="d1")
d1$Y2 <- d1$Y+2
gg2 <- ggplot(data=d1, aes(label=b)) +
  geom_point(aes(x=X, y=Y, col="blue")) +
  geom_line(aes(x=X, y=Y2, col="orange")) +
    scale_color_manual(values=c("blue","orange"),
                       labels=c("Left","Center"),
                       name="") 

print(gg2) results in:

But
ggplotly(gg2)

results in:

I know a more conventional code such as:

d1 <- data.frame(X=1:10, Y= (1:10)^2, b=paste0("B",1:10),position="left")
d2 <- data.frame(X=1:10, Y= (1:10)^2 + 2, b=paste0("B",1:10),position="center")
d <- rbind(d1,d2)
gg2 <- ggplot(data=d, aes(label=b)) +
  geom_point(data=d[d$position=="left",],aes(x=X, y=Y,col=position)) +
  geom_line(data=d[d$position=="center",],aes(x=X, y=Y,col=position)) +
  scale_color_manual(values=c("blue","orange","red"),
                     labels=c("Left","Center"),
                     name="") 
gg2
ggplotly(gg2)

results in similar legends, but I wonder if the 1st example is not uncovering a bug in ggplotly()

Commencez par exécuter l’exemple R reproductible fourni dans l’issue et comparez les légendes de ggplot2 et ggplotly. Étudiez la conversion par ggplotly des échelles de couleurs manuelles et vérifiez que la légende convertie correspond aux libellés déclarés Left et Center, sans entrées supplémentaires.