Set default button in rangeselector

Stumbled across this topic on SO and wasn't entirely satisfied with the suggested workaround. Is this something that you might want to have a look at?

Here's a minimum reprex: say I wanted to plot daily ozone readings from the airquality dataset against time and add range selectors,

library(plotly)

dat = airquality
dat$Date = as.Date(paste(1973, dat$Month, dat$Day, sep = "-"))

p0 = plot_ly(dat, x = ~ Date, y = ~ Ozone, type = "bar")

p1 = p0 %>% layout(
  xaxis = list(
    rangeselector = list(
      buttons = list(
        list(
          count = 1
          , label = "1 mo"
          , step = "month"
          , stepmode = "backward"
        )
        , list(step = "all")
      )
    )
  )
)
p1

, then the default zoom level would always be set to full range. If I wanted the default range to cover the last month only, I could certainly circumvent that through specifying a custom range as suggested in the linked answer,

p1 %>% layout(
  xaxis = list(
    range = c(
      {
        dt = max(dat$Date)
        lubridate::month(dt) = lubridate::month(dt) - 1
        dt
      }
      , max(dat$Date)
    )
  )
)

, which works nicely for point, line, etc. charts. Bar charts (not sure about other types), on the other hand, are being cut off at the first and last position. Now, I could also circumvent that by adding something like paste(dt, "10:00:00") and paste(max(dat$Date), "10:00:00") to my custom range, but I wonder whether this could be made a lot more user-friendly through enabling something like active = TRUE directly in the buttons definition?

Reproduziere das airquality-reprex mit plot_ly(), layout(), xaxis und rangeselector-Schaltflächen und vergleiche dann den Workaround für den benutzerdefinierten Bereich mit dem angeforderten Verhalten der Standardschaltfläche. Verfolge, wie die rangeselector-Schaltflächen und der xaxis-Bereich verarbeitet werden. Als erledigt gilt es, wenn eine Schaltfläche den initialen Bereich festlegen kann, ohne dass Balkendiagramme abgeschnitten werden, und das bestehende Beispiel weiterhin funktioniert.