Fail to use entry table with integer values

When CIGALE reads an input catalogue is replaces all the -9999 cells by NaN. When the input catalogue contains an error column containing -9999 everywhere, if this catalogue uses an ASCII format astropy things the column contains an integer which can not be replace by NaN. In that case, a CIGALE run stops:

ValueError: cannot convert float NaN to integer

It may have no meaning to have a column with only -9999 and to use it. Nevertheless, this can occur when processing a lot of catalogues with the same work-flow.

The problem occur whenever a flux or error column is interpreted as integer and CIGALE has to mask the incorrect values.