Model-based extraction:
- How does grizli specify parameters to vary?
- What is the merit-function for the fit?
- How are the data weighted ?
- What optimizer is used?
- How are uncertainties calculated?
- Which modules/functions/classes are doing the fitting?
- In model.GrismDisperser, do compute_model and compute_model_psf correspond to pixel- and object-based dispersion, respectively?
