Kfactors for compound datasets

Created by: giacomomagni

While trying to implement new k-factors files I found a potential bug. CF files with the new FK tables naming convention are not correctly loaded from the ymldb in case of compound observables. In particular in loader.py CF names should be read from 'operands' and not 'target_dataset'.

I propose the following fix:

        cfac_name = metadata["target_dataset"]
        # check whether there is a compound file
        cpath = theory.path / "compound" / f"FK_{cfac_name}-COMPOUND.dat"
        if cpath.exists():
            # Get the target filenames
            tnames = [i[3:-4] for i in cpath.read_text().split() if i.endswith(".dat")]
            cfactors = [self.check_cfactor(theoryID, i, cfac) for i in tnames]
        else:
            cfac_name = metadata["operands"]
            cfactors = []
            for operand in cfac_name:
                for fk_tab in operand:   
                    cfactors.append(self.check_cfactor(theoryID, fk_tab, cfac))
        ###

I suggest to deprecate or raise an error or a warning when the the folder compound is still present in a theory following the new naming convention, as this can lead to potential inconsistencies.