geom_image() for plotnine? Enhancement? With chessboard example

I would appreciate something like geom_image -library(ggimage)-
https://github.com/GuangchuangYu/ggimage
for plotnine.

geom_image allows to place an image at aes(x,y)

geom_image(
       mapping = NULL,
       data = NULL,
       stat = "identity",
       position = "identity",
       inherit.aes = TRUE,
       na.rm = FALSE, 
       by = "width",
       nudge_x = 0,
       ...
)

Not difficult to achieve? Here is an example using plt:
https://github.com/TerryGamon/Chess4Python

import pandas as pd
import plotnine as p9
import matplotlib.pyplot as plt
import matplotlib

def drawPosition(fen):
    figuren = pd.DataFrame({
    "f": ["p", "r", "n", "b", "q", "k", "P", "R", "N", "B", "Q", "K"],
    "gif": [
        "bp1024.gif", "br1024.gif", "bn1024.gif", "bb1024.gif", "bq1024.gif", 
        "bk1024.gif", "wP1024.gif", "wR1024.gif", "wN1024.gif", "wB1024.gif", 
        "wQ1024.gif", "wK1024.gif"
        ]
    })

def square_color(s, z):
    return "black" if (s + z) % 2 == 0 else "white"

def fen_to_dataframe(fen):
    board_fen = fen.split()[0]
    board = []
    rows = board_fen.split('/')
    for z, row in enumerate(rows):
        s = 1  
        for char in row:
            if char.isdigit():
                for _ in range(int(char)):
                    color = square_color(s, 8 - z)
                    board.append({'s': s, 'z': 8 - z, 'f': None, 'c': color})
                    s += 1
            else:
                color = square_color(s, 8 - z)
                board.append({'s': s, 'z': 8 - z, 'f': char, 'c': color})
                s += 1
    
    df = pd.DataFrame(board)
    return df

df_board = fen_to_dataframe(fen).merge(figuren, on='f', how='left')

p=(p9.ggplot(df_board)
+p9.geom_tile(p9.aes(x='s-.5',y='z-.5', fill='c'))
#+p9.geom_text(p9.aes(x='s-.5',y='z-.5', label='f'), size=15)
+p9.theme_bw()
+p9.theme(figure_size=[8,8])
+p9.coord_fixed()
+p9.scale_fill_manual(values=['darkgray','lightgray'])
+p9.theme(axis_line=p9.element_blank())
+p9.theme(panel_background=p9.element_blank())
+p9.theme(axis_text=p9.element_blank())
+p9.theme(axis_ticks=p9.element_blank())
+p9.theme(panel_grid=p9.element_blank())
+p9.labs(x='',y='')
+p9.theme(legend_position='none')
)

fig = p.draw()
ax = fig.axes[0]


dff_board = df_board.dropna()
# here geom_image happens
for i, row in dff_board.iterrows():
    img = matplotlib.offsetbox.OffsetImage(
        plt.imread(row['gif']), 
        zoom = 0.06)
    ax.add_artist(matplotlib.offsetbox.AnnotationBbox(img,
        (row['s']-.5 ,row['z']-.5)
        ,pad = 0.0
        ,frameon =False
        )
        )
return fig

drawPosition("r1b1k1nr/p2p1pNp/n2B4/1p1NP2P/6P1/3P1Q2/P1P1K3/q5b1")

Start by reviewing plotnine's existing geom_tile usage and the matplotlib OffsetImage and AnnotationBbox example shown in the issue. Define the feature's scope and rendering behavior for placing images at aesthetic x and y positions, then add a chessboard-style example demonstrating the completed geom_image equivalent.