Close #22

Handle the units for properties in log. Note that currently they cannot be saved properly in FITS files due to the format limit.

Switch the implementation to a suggestion made by Yannick. The speed is similar for small catalogues but it is substantially faster for large catalogues.

Handle the case where some bands do not have a best fit (e.g., in case the flux was not calculated because no object has a flux and the user did not add this band to the "band" parameter in pdf_analysis).

Ensure the figure is properly created and minimise the disturbance caused by the logo by placing it in the lower left corner.

Eliminate the unphysical wavy structure of the nebular continuum in the mid-infrared by pre-interpolating it on a refined wavelength grid in log-log space.

Improve the sanitation of input data so that upper limits of extensive properties are not eliminated from the quantities to be fitted when upper limits are activated

For some reason assigning an integer array to np.nan, which should never be possible, only fails when it is of size 1. Work around this with a try/except.

Allow to save the χ² only related to fluxes or to properties rather than an all or nothing approach.

Get back the modifications made to generate the wheels (and to remove
obsolete version numbers).