Pandas version seems to matter

[nheeren]

Created a new virtual environment today with the brand new pandas 0.24.0. It seems like it includes some subtle breaking changes. For instances, in read_candidate_meta() read_excel() now assumes the names kwargs includes a name for the index column. Before this was not the case.

One (me) should check:

• backward compatibility of the current / future code
• if necessary define a requirement

P.S.: Given the number of bugfixes on read_excel(), it is also quite possible that my code was before "using" a bug which is now fixed. Need to explore further...