Keep the MultiIndex of the df even if empty

[michael-hoss]

I just faced this:

• In the given line, copy.index was an empty pd.MultiIndex with specified columns ['FrameId', 'Event'], as it was supposed to be.
• Then, however, the line turned it into an empty Index (not MultiIndex) without any specified columns.
• This messed up r further on: its index also became just Index, though, with a tuple in each entry. This screwed up r.index.get_level_values(0).max() + 1 because you cannot add 1 to a tuple.

The proposed change should prevent this from happening while maintaining the intended behavior.

I am using pandas==2.1.2 and Python 3.10

[cheind]

Hey thanks for the PR. I'm running CI pipeline to ensure tests pass.

[cheind]

thanks, merged

[michael-hoss]

Awesome, thanks for the quick reaction