Aspect ratio with small multiples / subplots

I would like to create small multiples with a fixed aspect ratio. However, there seems to be a but somewhere and only some plots retain the correct aspect ratio.
From what I gathered, setting a fixed aspect ratio with plotly is done as follows: layout(yaxis = list(scaleanchor = "x"))

Consider the follwing example:

df <- data_frame(
  x = rep(1:5,25),
  y = rep(1:5,25),
  g = sort(rep(1:25,5))
)

plots <- df %>%
  split(.$g) %>%
  map(function(x){
    plot_ly(data = x, x = ~x, y = ~y, type = "scatter",mode = "lines") %>% add_lines()
  })

small_mulitples <- subplot(plots,nrows = 5) %>%
  layout(yaxis = list(scaleanchor  = "x")) %>%
  hide_legend()

If I plot small_mulitples, only the first plot (row 1, column 1) has an aspect ratio of 1. The other have an arbitrary aspect ratio.

Here's the link to the interactive version of this plot: https://plot.ly/~rata_zhaw/1/

Interestingly, if I choose the option shareX = T in subplot(), the entire first column has the correct aspect ratio. Nothing changes if I choose shareY = T

small_mulitples <- subplot(plots,nrows = 5, shareX = T) %>%
  layout(yaxis = list(scaleanchor  = "x")) %>%
  hide_legend()

And here's the link to the interactive version of the second plot: https://plot.ly/~rata_zhaw/3/

If I plot any of the plots individually, the aspect ratio is correct:

plots[[10]] %>%
  layout(yaxis = list(scaleanchor  = "x")) %>%
  hide_legend()

Start by reproducing the supplied R example with subplot(), nrows = 5, and layout(yaxis = list(scaleanchor = "x")); compare it with the shareX = T and single-plot cases. Trace how subplot() combines the plot layouts, and consider the issue done when every small multiple retains the fixed aspect ratio without requiring shared axes.