Problem with radialaxis ticks in polar plot when using date values

Hello, I managed to create a polar heatmap plot following the examples given in #2024, but when trying to use time as the radial dimension I found some problems with the ticks. I tried with tickmodes auto, linear and array but got wrong ticks with all of them. The time values shown in the hover are correct, only the ticks appear to be wrong. Does anyone have a suggestion on what may be the problem?
Thank you.

import numpy as np
import pandas as pd
import plotly.graph_objects as go
from plotly.offline import plot

from datetime import datetime, timedelta

# dt_start, dt_end, dt_step ='2000-01-01', '2001-01-01', '1MS'
dt_start, dt_end, dt_step ='2000-01-01', '2000-01-10', '1D'

times = pd.date_range(dt_start, dt_end, freq=dt_step).values

dirs = np.arange(180, 271, 10)

r, theta = np.meshgrid(times, dirs)

z = np.random.random(r.shape)

hovertemplate = (f'date: %{{r|%Y-%m-%d %H:%M:%SZ}}<br>'
                 f'direction: %{{theta}}<br>'
                 f'value: %{{customdata[0]:.2f}}'
                 '<extra></extra>')

fig = go.Figure(go.Barpolar(r=r.ravel(),
                            theta=theta.ravel(),
                            marker_color=z.ravel(),
                            customdata=np.expand_dims(z.ravel(), 0).T,
                            hovertemplate=hovertemplate))

fig.update_layout(polar=dict(
    angularaxis=dict(tickvals=np.arange(0, 360, 45),
                     direction='clockwise'),
    radialaxis=dict(angle=45,
                    tickangle=45,
                    tickfont={'size': 14},
                    tickformat='%Y-%m-%d %H:%M:%S',

                    # tickmode='auto',
                    # nticks=5,

                    # tickmode='linear',
                    # tick0=times[0],
                    # dtick='1D',

                    tickmode='array',
                    tickvals=times,

                    )))

fig.show()

With tickmode='auto':

With tickmode='linear':

With tickmode='array':

Using plotly.version = '5.4.0'

Commencez par exécuter l’exemple Python fourni avec Plotly 5.4.0 et comparez les graduations de l’axe radial produites par les modes auto, linear et array. Suivez la gestion des graduations de l’axe radial pour les données r contenant des valeurs de date. Le travail est terminé lorsque les trois modes pris en charge produisent des graduations de date correctement positionnées et formatées.