Improved error inflation, new log-normal cross sections, and some minor things

[tomasstolker]

• Improved approach for fitting the error inflation of spectra in FitModel. Instead of fitting a constant, the free parameter is a scaling relative to the model fluxes (see Piette & Madhusudhan 2020). The use of the error inflation parameter has not changed but in this case typically requires a bound value between 0 and 1. Both run_multinest and run_mcmc have been updated.
• The optional model parameter has been added to update_spectra since the error inflation requires calculating the model spectrum.
• Bug fix with the use of fixed parameters in FitModel since the calibration and extinction parameters were not fixed.
• The deredden parameter in add_object can also be a float (instead of dictionary), in which case the same dereddening is applied to all input data.
• When fitting an atmosphere + disk, the disk_teff and disk_radius parameters are included in the luminosity calculation of plot_posterior.
• A new grid of dust cross sections with a log-normal distribution has been calculated with a small improvement in the sampling of the distribution and in this case also ignoring grains smaller than 1 nm. The plot_size_distributions function has been updated accordingly. The new cross sections are downloaded and added to the Database with add_dust.
• Updated README, documentation, and code comments.