px.scatter_3d, use case where color param affects size output

I have created a scatter3d chart with plotly express. The scatter3d works fine with column "count" as size (fig 1). There are three big bubbles on the top layer and one on the bottom layer.

fig1

I next add the color parameter to be a column "type" in the frame. The resulting plot appears to have incorrectly swapped out "count" for "type" in the size parameter, so now the actual size param is ignored and it maps color to size. (fig 2).

fig2

import pandas as pd
import plotly.express as px

df = pd.DataFrame({
    "campaign": ["funny", "funny", "funny", "funny", "patriotic", "patriotic", "patriotic", "patriotic"],
    "product": ["classic", "classic", "diet", "diet", "classic", "classic", "diet", "diet"],
    "type": ["click", "land", "click", "land", "click", "land", "click", "land"],
    "count": [397, 108, 108, 411, 99, 410, 144, 323]
})

fig1 = px.scatter_3d(df, x="campaign", y="product", z="type", hover_data="count").update_traces({"marker": {"size": [j/10 for j in df["count"]]}})

fig2 = px.scatter_3d(df, x="campaign", y="product", z="type", hover_data="count", color="type").update_traces({"marker": {"size": [j/10 for j in df["count"]]}})

Start by running the supplied pandas and plotly.express px.scatter_3d reproducer, comparing fig1 and fig2 after update_traces. Inspect the generated marker size and color values to find why adding color changes the size output. Done means the count-derived marker sizes remain effective when color="type" is supplied.