customdata parsed incorrectly when array contains more than 1 column in piechart

When using more than one column to define customdata, the data seems to be parsed as a tuple.
The example below shows three versions of the same figure:

• LEFT. Only one column of customdata. Works as expected.
• CENTER. Two columns of customdata. The first column in customdata is not parsed correctly when using a :.1s format. The second column in customdata is not parsed correctly by hovertemplate in go.Pie but is parsed correctly by texttemplate.
• RIGHT. Workaround using customdata as a tuple and selecting its first %{customdata[0][0]} and second %{customdata[0][1]} values accordingly.

import pandas as pd
import plotly.graph_objects as go
from plotly.subplots import make_subplots

df = pd.DataFrame({
    'country': ['A', 'B', 'C'],
    'count':   [10, 20, 30],
    'total': [1.5e5, 4.5e4, 3.2e3],
    'ratio': [0.05, 0.10, 0.15]
})

fig = make_subplots(rows=1, cols=3, specs=[[{"type": "domain"}, {"type": "domain"}, {"type": "domain"}]])

# customdata parsed correctly with only one column
fig.add_trace(go.Pie(
    labels=df['country'],
    values=df['count'],
    customdata=df[['total']], # 1 column
    hovertemplate=(
        'Country: %{label}<br>'
        'Count: %{value}<br>'
        'Total documents: %{customdata[0]:.1s}'),
    hole=0.3
),
row=1, col=1)

# Problem with customdata
fig.add_trace(go.Pie(
    labels=df['country'],
    values=df['count'],
    customdata=df[['total', 'ratio']], # >1 columns
    hovertemplate=(
        'Country: %{label}<br>'
        'Count: %{value}<br>'
        'Total documents: %{customdata[0]:.1s}<br>'
        'Ratio: %{customdata[1]:.2f}'),
    texttemplate = "%{customdata[1]:.2p}", # correctly parsed
    hole=0.3
),
row=1, col=2)

# Workaround
fig.add_trace(go.Pie(
    labels=df['country'],
    values=df['count'],
    customdata=df[['total', 'ratio']], # >1 columns
    hovertemplate=(
        'Country: %{label}<br>'
        'Count: %{value}<br>'
        'Total documents: %{customdata[0][0]:.1s}<br>' # as tuple
        'Ratio: %{customdata[0][1]:.2f}'),
    texttemplate = "%{customdata[1]:.2p}",  # correctly parsed
    hole=0.3
),
row=1, col=3)

fig.show()

Tested using:
Plotly.py: 6.0.1
Pandas: 2.1.4

Start by running the supplied Plotly.py and Pandas reproduction, then trace how multi-column customdata is passed to pie-chart hovertemplate formatting. Done means both customdata columns render with the requested formats without tuple-style indexing, while the one-column case and texttemplate behavior remain correct.