Bar charts don't group when there's only 1 group

[logankopas]

I found this issue while trying to work around #5294

Plotly version: 6.2.0

Minimal Example

import plotly.graph_objects as go
import pandas as pd
import datetime

data = [
    {"date": datetime.date(2025,7,20), "net_sales": 23.34, "gross_sales": 32.98},
    {"date": datetime.date(2025,7,21), "net_sales": 45.21, "gross_sales": 68.12},
    {"date": datetime.date(2025,7,22), "net_sales": 12.98, "gross_sales": 34.54},
    {"date": datetime.date(2025,7,23), "net_sales": 41.32, "gross_sales": 45.32},
]

df = pd.DataFrame.from_records(data)

def make_figure(df, width=None):
    fig = go.Figure()
    fig.add_trace(go.Bar(
        x=df['date'],
        y=df['gross_sales'],
        xperiod=24*60*60*1000,
        xperiod0=df['date'].min(),
        xperiodalignment="middle",
        width=24*60*60*1000 * 0.4,
        opacity=0.6,
        offsetgroup='gross'  # same behaviour exhibited whether or not this is specified
    ))
    fig.add_trace(go.Bar(
        x=df['date'],
        y=df['net_sales'],
        xperiod=24*60*60*1000,
        xperiod0=df['date'].min(),
        xperiodalignment="middle",
        width=24*60*60*1000 * 0.4,
        opacity=0.6,
        offsetgroup='net'  # same behaviour exhibited whether or not this is specified
    ))
    fig.update_layout(
        barmode='group',
        xaxis={
            'ticklabelmode': 'period',
            'tickmode': 'auto',
            'dtick': 24*60*60*1000,
            'range': (df['date'].min(), df['date'].max() + datetime.timedelta(days=1)),
        })
    fig.show(renderer="browser")

make_figure(df)
make_figure(df[0:1])

When plotting a grouped bar chart of aggregated timeseries data, strange behaviour happens when there's only 1 group in the data. When there are multiple groups of data each bar is displayed as one would expect (first screenshot below), but when there's only a single data point the bars overlap (second screenshot). These bars overlap regardless of whether the offsetgroup or alignmentgroup is specified, and the offset must be manually set in order to get consistent behaviour regardless of the data provided.

[gvwilson]

@camdecoster please check if this is happening at the JS level.