Bubble changing color and representing a different category in animated scatter plot

I was trying to create an animated scatter plot with each bubble representing a race category indicated by a color. In the actual data set, some race categories don't data in certain years. When those years were excluded, the plot looked fine. However, when those years were not excluded, some bubbles changed colors and turned into a different race category . For example, in the code below, race 4 changed into race 5 in year 6 when race 4 actually does not have data in year 6 - year 9. Thanks!

str <-
  "seq, race, achievement, growth, size
year 1, race 6, 41.5, 35.5,  54
year 2, race 6, 45.5, 48.5,  54
year 3, race 6, 37.0, 34.5,  64
year 4, race 6, 42.0, 29.5,  62
year 5, race 6, 39.0, 40.0,  60
year 6, race 6, 39.0, 41.0,  71
year 7, race 6, 33.0, 40.0,  74
year 8, race 6, 25.0, 42.0,  59
year 9, race 6, 24.0, 55.0,  59
year 1, race 2, 22.0, 36.0, 357
year 2, race 2, 23.0, 43.0, 357
year 3, race 2, 21.0, 42.0, 359
year 4, race 2, 18.0, 45.0, 364
year 5, race 2, 21.0, 42.0, 351
year 6, race 2, 20.0, 54.0, 353
year 7, race 2, 17.0, 41.0, 370
year 8, race 2, 13.0, 46.0, 319
year 9, race 2, 14.0, 48.0, 318
year 1, race 3, 39.0, 47.0,   2
year 2, race 3, 29.5, 28.5,   2
year 3, race 3, 36.0, 29.5,   2
year 4, race 3, 34.5, 64.5,   2
year 5, race 3, 35.5, 70.5,   2
year 6, race 3, 69.0, 35.0,   1
year 7, race 3, 60.0, 21.0,   1
year 1, race 5, 65.0, 59.0,  23
year 2, race 5, 68.0, 44.0,  23
year 3, race 5, 54.0, 57.0,  17
year 4, race 5, 49.0, 43.0,  19
year 5, race 5, 45.5, 49.0,  20
year 6, race 5, 54.0, 64.0,  23
year 7, race 5, 48.0, 56.0,  32
year 8, race 5, 37.0, 68.0,  35
year 9, race 5, 37.0, 58.0,  37
year 1, race 7, 35.0, 46.0,  99
year 2, race 7, 40.0, 57.5, 100
year 3, race 7, 37.0, 40.0, 113
year 4, race 7, 36.0, 51.5, 110
year 5, race 7, 33.0, 41.0, 108
year 6, race 7, 39.0, 53.0, 101
year 7, race 7, 30.0, 40.0, 106
year 8, race 7, 28.0, 53.0, 111
year 9, race 7, 28.0, 53.0, 106
year 1, race 4, 74.5, 93.0,   2
year 2, race 4, 69.5, 80.5,   2
year 3, race 4, 84.0, 98.0,   1
year 4, race 4, 82.0, 44.0,   1
year 5, race 4, 60.0,  3.0,   1
year 1, race 1, 47.0, 38.0, 461
year 2, race 1, 48.0, 42.5, 454
year 3, race 1, 49.0, 41.0, 449
year 4, race 1, 48.0, 52.0, 461
year 5, race 1, 47.0, 42.0, 448
year 6, race 1, 52.0, 47.0, 412
year 7, race 1, 47.0, 39.0, 416
year 8, race 1, 38.0, 46.5, 362
year 9, race 1, 41.0, 51.0, 362"
dt <- read.table(textConnection(str), header=T, sep=",", na.strings=" ")

plot_ly(dt) %>% 
  add_markers(x = ~achievement, y = ~growth, size = ~size, 
              color = ~race, 
              frame = ~seq,
              hoverinfo = 'text', text = ~paste(race, size, sep = '\n'), 
              marker = list(opacity = 0.5, sizemode = 'diameter'))  

Empieza ejecutando el ejemplo reproducible de R del issue con plotly.R y compara los frames animados en los que la carrera 4 no tiene datos. Traza cómo se asignan los valores de los frames a los colores de los marcadores. La tarea estará terminada cuando cada carrera conserve el color que tiene asignado cuando faltan categorías en años posteriores.