Fixed monomer pickling to avoid name clashes. This breaks pickle backwards-compatiblity! [see #35] Previosly all monomer classes were stored a single namespace, allpy.data.monomers. This caused a few name clashes, which were mostly resolved, and one name clash, that was not. (This caused one class to be named differently depending on the order in which modules were loaded). Now, instead of one allpy.data.monomers module we have allpy.data.monomers package with modules dna, rna, protein. This ensures that all name clashes are resolved uniformly upon any sequence of modules loading. This may also help in future to keep backward-compatibility longer in case we replace dynaminc monomer class creation with storing the classes in the module - if we want to retain independent loading of dna/rna/protein parts.