Choropleth maps only render the first feature with a matching featureidkey, not all matching features

When px.choropleth() is called with a GeoJSON feature collection containing multiple features matching a single featureidkey value, only the first feature is rendered. I believe all features should be so rendered.

The following code generates two images: the first renders all four features, while the second renders only three. The first uses a featureidkey unique to all features, the second uses a featureidkey that is not unique -- only the first feature to match is rendered. :-(

#!/usr/bin/env python3

import pandas as pd
import plotly.express as px

# A simple GeoJSON with four features.  
# Each feature has a district.  
# One district has two features.
geojson = {
    "type": "FeatureCollection",
    "features": [
        {
            "type": "Feature",
            "properties": {
                "name": "Alpha",
                "district": "One",
            },
            "geometry": {
                "type": "Polygon",
                "coordinates": [[[0, 0], [0, 5], [5, 5], [5, 0], [0, 0]]],
            },
        },
        {
            "type": "Feature",
            "properties": {
                "name": "Bravo",
                "district": "One",
            },
            "geometry": {
                "type": "Polygon",
                "coordinates": [[[5, 0], [5, 5], [10, 5], [10, 0], [5, 0]]],
            },
        },
        {
            "type": "Feature",
            "properties": {
                "name": "Charlie",
                "district": "Two",
            },
            "geometry": {
                "type": "Polygon",
                "coordinates": [[[0, 5], [0, 10], [5, 10], [5, 5], [0, 5]]],
            },
        },
        {
            "type": "Feature",
            "properties": {
                "name": "Delta",
                "district": "Three",
            },
            "geometry": {
                "type": "Polygon",
                "coordinates": [[[5, 5], [5, 10], [10, 10], [10, 5], [5, 5]]],
            },
        },
    ],
}

districts = {
    f["properties"]["name"]: f["properties"]["district"] for f in geojson["features"]
}

# A simple dataframe relating names and districts from the above GeoJSON data.
# i.e., [{'name': 'Alpha', 'district': 'One'}, ...]
data = [f['properties'] for f in geojson['features']]

df = pd.DataFrame(data)

# This code generates two maps from the data:
# * The name map shows all four squares in the same color (count = 1)
# * The district map shows _three_ squares, one with one color (count = 2) and two with another color (count = 1)

# What I expected:
# * The district map should show all four squares, two with light color and two with dark
for key in ["name", "district"]:

    region_counts = df.groupby([key], observed=False).size().reset_index(name="count")

    fig = px.choropleth(
        region_counts,
        geojson=geojson,
        fitbounds='geojson',
        locations=key,
        color="count",
        featureidkey=f"properties.{key}",
    )

    fig.write_image(f"mwe-{key}.png")

Reproduce the two px.choropleth() calls from the supplied script, comparing the unique name and repeated district values through locations and featureidkey. Trace the choropleth entry point and GeoJSON feature matching, then verify completion by rendering all four polygons in the district map while preserving the unique-name result.