Dash 2.0 DataTable Previous Data Loses Column Order

In dash v1.19.0 with dash-table v4.11.2 the data_previous property of a DataTable retained the order of the columns
In dash v2.0.0 this property now returns the columns ordered alphabetically; not based on previous ordering

Minimal reproducible example:

from dash import Dash, Input, Output, State
from dash.exceptions import PreventUpdate
from dash.html import Div, Button
from dash.dash_table import DataTable
from pandas import DataFrame

df = DataFrame(dict(Z=[1, 2, 3], A=[7, 8, 9], L=[4, 5, 6]))
app = Dash(__name__)
table = DataTable(id='table', columns=[dict(name=i, id=i, deletable=True) for i in df.columns],
                  data=df.to_dict('records'))
undo_button = Button(id='undo-button', children='Undo')
app.layout = Div([Div(table), Div(undo_button)])


@app.callback([Output(table.id, 'columns'), Output(table.id, 'data')],
              Input(undo_button.id, 'n_clicks'),
              State(table.id, 'data_previous'))
def undo(n_clicks, previous_data):
    if n_clicks is None:
        raise PreventUpdate
    df = DataFrame(previous_data)
    return [dict(name=i, id=i, deletable=True) for i in df.columns], df.to_dict('records')

if __name__ == '__main__':
    app.run_server()

If I delete any of these columns, and then click the undo button, I expect the deleted column to be restored, and the order of the columns to remain Z, A, L

Initial table:

Deleted column:

Following undo:

Run the minimal reproducible example with the stated Dash and dash-table versions, focusing on DataTable's data_previous property after deleting a column. Compare the returned record order between versions and verify that restoring the deleted column preserves the original Z, A, L order.