Data frame fails when columns are integers

We should convert column names to character in the decorator.

from shiny import App, reactive, ui, render
import pandas as pd

df = pd.DataFrame({1: [1, 2, 3], 2: [4, 5, 6]})

app_ui = ui.page_fluid(ui.output_data_frame("char_df"), ui.output_data_frame("int_df"))


def server(input, output, session):
    @output
    @render.data_frame()
    def char_df():
        df2 = df.copy()
        df2.columns = df2.columns.astype(str)
        return df2

    @output
    @render.data_frame()
    def int_df():
        return df


app = App(app_ui, server, debug=False)

Start with the render.data_frame decorator and reproduce the example using the integer-named pandas DataFrame. Compare it with the version whose columns are converted to strings; done means both data frames render successfully with equivalent column labels.