to_dataframe fails if dataarray has dimension 1

The to_dataframe method fails with ValueError if the dataarray has only value

MCVE Code Sample

# Your code here
x = np.arange(10)
y = np.arange(10)

data = np.zeros((len(x), len(y)))

da = xr.DataArray(data, coords=[x, y], dims=['x', 'y'])

da.sel(x=1,y=1).to_dataframe(name='test')

Expected Output

Expect a dataframe with one row

Problem Description

This happened when selecting a single value out of a gridded dataset - in cases where there was only one value output the to_dataframe failed.

Output of xr.show_versions()

Start at the DataArray.to_dataframe entry point and reproduce the provided MCVE, focusing on the scalar DataArray produced by selecting x=1 and y=1. Add regression coverage for this case and verify that the result is a one-row dataframe without raising ValueError.