query dplyr groups from data only if the relevant trace has a notion …

[cpsievert]

Closes #1351

TESTING NOTES

Run the following and make sure it produces the image below. Make sure the point labelled "1" (not "3") is black.

# devtools::install_github("ropensci/plotly#1386")
library(plotly)

mtcars %>%
    mutate(id = seq_len(nrow(.))) %>%
    group_by(cyl) %>%
    plot_ly(
      x = ~mpg, y = ~wt, text = ~id,
      marker = list(color = ~ifelse(id == 1, "black", "red")),
      mode = "marker+text", textposition = "right"
    )

[cpsievert]

This chunk was moved here from lines 150-154, so we can be sure we know the trace type before applying has_groups(trace)