Special bindings for stars data structures

A good start would be to replicate the plot.stars() method (from https://github.com/r-spatial/stars)

library(stars)
etm <- st_stars(system.file("tif/L7_ETMs.tif", package = "stars"))
plot(etm)

So plot_ly(etm) would do something similar to this (automatically)

library(sf)
etm <- st_transform(etm, 3857)
band <- st_dimensions(etm)[["band"]]
n <- band$to - band$from + 1
ann <- list(x = 0.5, y = 1, yanchor = "bottom", xanchor = "middle", xref = "paper", yref = "paper", showarrow = FALSE)
plots <- lapply(seq_len(n), function(i) {
  z <- t(etm[[1]][,,i])
  ann$text <- paste(names(etm), i)
  plot_ly(z = z[seq.int(nrow(z), 1), ]) %>%
    layout(annotations = ann) %>%
    colorbar(title = i)
})
subplot(plots, nrows = 2, titleX = TRUE, shareX = TRUE, shareY = TRUE)

cc @jackparmer

Start by studying the referenced stars::plot.stars() behavior and the shown plot_ly(etm) and subplot() examples. Identify the existing plot_ly and subplot entry points that handle structured data, then define completion as automatic plotting of stars objects with band panels and matching annotations, colorbars, and shared axes.