This is a somewhat incomplete explanation based on a quick scan of the source code, but from what I can tell, the fastsaxs scoreterm only looks at protein residues. Any non-protein residues (like nucleic acids, ligands or metal atoms) are ignored.
The saxs_score term is a Centroid one, and is really specific for protein residues as well. It doesn’t have all the parameters needed to work on nucleic acids or metal ions.
There is also a saxs_fa_score term, which is an FA version of the saxs_score term. This is a bit more general than the fastsaxs term, in that some of the nucleic acid atoms are counted, but some of the atom types which are present for nucleic acids but not protein residues (most noticiably the phosphorus) are missing and will effectively be “invisible” to the saxs calculation. (This also includes any metal ions present in the structure.)