[BUG]: deserialization of xrange of box select in figure with multiple subplots broken

Description

When you do a box (or lasso) select in a figure with multiple subplots (not the first subplot) the selection x/yrange is not a list/dict but a plain python object.

The gif above was created with this example:

import plotly.graph_objects as go
import solara
from plotly.subplots import make_subplots

def build_figure() -> go.Figure:
	fig = make_subplots(rows=1, cols=2, subplot_titles=("Scatter", "Bar"))
	fig.add_trace(go.Scatter(x=[1, 2, 3, 4, 5, 6], y=[2, 1, 3, 5, 4, 6], name="Series A"), row=1, col=1)
	fig.add_trace(go.Bar(x=["A", "B", "C"], y=[3, 1, 2], name="Series B"), row=1, col=2)
	fig.update_layout(title="FigureWidget example with 2 subplots",)
	return fig

@solara.component
def Page():
	def on_selection(data):
		print("on_select data:", data, flush=True)
	solara.FigurePlotly(build_figure(), on_selection=on_selection)

Steps to reproduce

1. Create a figure with more than one subplot.
2. Do a box select in the second subplot.
3. Observe the data send to on_select.

Start by running the supplied Python example with plotly.subplots and solara.FigurePlotly, then inspect the on_selection callback data for a box selection in the second subplot. Trace how the selection x/yrange is deserialized for non-first subplots. Done means those ranges arrive as a list or dictionary rather than a plain Python object.