Refactor Uniprot sequence fetcher

There is room for improvement/optimisation of Uniprot sequence fetcher.
  I noticed that in order to construct an alignment the Uniprot sequence fetcher follows the process below after downloading a Uniprot xml and creating a UniprotEntry POJO from it:
   - A Fasta StringBuffer containing the name and residues is created via constructSequenceFastaHeader(...)
   - Then the StringBuffer is converted to String and parsed to a SequenceI instance via FormatAdapter().readFile(...);
   - Finally, addUniprotXrefs(...) is called to add DBRefEntries to the retrieved Sequence(s)
   - Also, I noticed dbrefs are stored as PDBRefEntry and later converted to DBRefEntry within the addUniprotXrefs(...) method

I suggest removing the methods constructSequenceFastaHeader(...) and addUniprotXrefs(...), and introducing a method toSequence() in UniportEntry which will create a SequenceI instance on the fly from the UniprotEntry members