File format independent data loading

[gwirn]

!! This PR introduces breaking changes !!
It is not possible any more to load multiple trajectories with multiple consecutive calls to PDBData.import_pdb()

With this PR MDAnalysis is used to read in one ore multiple trajectory files. This is now done with a single call:

• for one file without a topology file
  • PDBData.import_pdb("path/to/trajectory.pdb")
• for one file with a topology file
  • PDBData.import_pdb("path/to/trajectory.dcd", "path/to/topology.pdb")
• for multiple files without a topology file
  • PDBData.import_pdb(["path/to/trajectory0.pdb", "path/to/trajectory1.pdb"])
• for multiple files with a topology file
  • PDBData.import_pdb(["path/to/trajectory0.dcd", "path/to/trajectory1.dcd"], "topology.pdb")

It still preserves the original functionality of the PDBData class.

Additionally the atom selection has changed - it can still be done via a list of atoms like ["CA", "C", "N", "CB", "O"] but can now also use the MDAnalysis selection language

[degiacom]

Very useful update. It will be breaking for anyone previously loading several multiPDB datasets in succession, but the way to update such older code to conform with this new implementation is trivial (just put all the files in a single list passed as parameter).