Atoms in the fragments in the configs are again ordered as first labeled during Bond structure recognition
Due to the move to disconnected subgraph recognition an additional "layer" of molecules is inserted. The Atoms are AddCopyAtom()ed into new molecules representing each subgraph and receiving new labels thereby (even if there is just one subgraph in the end). As these labels were used in the sorted "KeySet" sets for adding atoms on the one hand, the KeySet file on the other hand however was created from the molecules directly now with labels from the topmost molecule structure ... : lines in BondFragment configs were swapped with respect to freshly fragmented and parsed from KeySet files.
Do ameloriate this, the StoreFragmentFromKeySet procedure was taken out of FragmentBOSSANOVA() and up into FragmentMolecule(), where the uppermost molecular structure is accessible. At the same time, the topmost labels (via GetTrueFather) are used in the KeySets. Thus, we receive again ordered list of atoms in each fragment config and they match perfectly with those re-created from parsed keyset files (check with diff over all BondFragment*.conf*).
