[BUG] LocIndexer of a DataFrame with a MultiIndex columns with a single label tuple returns DataFrame

[mroeschke]

Describe the bug
LocIndexer of a DataFrame with a MultiIndex columns with a single label tuple returns DataFrame instead of a Series

Steps/Code to reproduce bug

In [50]: import cudf

In [51]: cudf.__version__
Out[51]: '23.10.00'

In [52]: dfmi = cudf.DataFrame([list('abcd'),
    ...: 
    ...:                      list('efgh'),
    ...: 
    ...:                      list('ijkl'),
    ...: 
    ...:                      list('mnop')],
    ...: 
    ...:                     columns=cudf.MultiIndex.from_product([['one', 'two'],
    ...: 
    ...:                                                         ['first', 'second']]))

In [53]: dfmi.loc[:, ('one', 'second')]
Out[53]: 
  (one, second)
0             b
1             f
2             j
3             n

In [54]: dfmi.to_pandas().loc[:, ('one', 'second')]
Out[54]: 
0    b
1    f
2    j
3    n
Name: (one, second), dtype: object

Expected behavior
Like pandas, a single tuple label should be treated like a scalar label instead of a container of labels and return a Series instead of a DataFrame

Environment overview (please complete the following information)

• Environment location: Bare-metal
• Method of cuDF install: conda
  • If method of install is [Docker], provide docker pull & docker run commands used

Environment details
Please run and paste the output of the cudf/print_env.sh script here, to gather any other relevant environment details

Additional context
Add any other context about the problem here.