Refactor mapping for sequence <-> (PDB resnum <-> PDB atom position)

Currently this mapping done using a two dimensional int array. Using a two dimensional array for this type of mapping introduces a lot of unnecessary redundancy and doesn't capture some corner case.

I propose we use a HashMap <Integer, int[]> where the hash key is the residue index and the value is an int array containing the PDB resnum index, and PDB atom position for the given residue.

This model will not only eliminate the unnecessary redundancies, but will equally capture most corner cases more accurately and increase performance (HashMap performs better than Arrays).