This was a necessary evil since the upcoming PM3 release relies on having the
newly added sec_struct functionality in place. The changes are additive and do
not break any existing code based on previous OST versions.

The reason is that we iterate over the atoms and add the per atom
accessibility to the current value. You can guess what happens when
calling the Accessibility function several times...

structure of single residues

This changes behaviors in MatchResidueByLocalAln and MatchResidueByGlobalAln (
or Superpose with match = local-aln or global-aln) which previously failed for
non-"protein" chains (i.e. anything w/o peptide bonds). Now, amino acid chains
without peptide bonds (e.g. CA-only) are acceptable and other chains (e.g.
ligands / water) are skipped correctly.