xAODPrimitives Python Helper, master branch (2020.05.27.)

This is meant to fix the issue(s) described in ATR-21477.

As @fwinkl reported, unfortunately the xAODCoreRflxDict -> xAODCoreDict name change introduced with !33176 (merged) caused a number of failures in the last nightly.

Instead of just changing the names of the dictionary libraries loaded in our Python configuration explicitly, I decided to do something better. I introduced some python code into xAODPrimitives that could be used to provide access to the enumerations defined in the xAOD::Iso namespace. Instead of clients loading libraries explicitly left and right. (As it turns out, xAODCoreRflxDict was always loaded to get access to those enumerations. Which by itself is a weird thing...)

While making changes in the clients so that they would use the new xAODPrimitives.xAODIso module, I also made some other changes.

• Reconstruction/egamma/egammaTools/python/EMShowerBuilder.py didn't apparently need any of these enumerations by now, the code could just be removed;
• PhysicsAnalysis/SUSYPhys/LongLivedParticleDPDMaker/python/HNLFlags.py:
  • Made sure that the module could be loaded by itself. (Previously it couldn't be, as I found in my testing.)
  • Removed the _v1 postfixes from the xAOD::Muon type names, as with the recently introduced typedefs ROOT.xAOD.Muon is now a known type.

These cover every single place where xAODCoreRflxDict was mentioned in our code. So to speed things up, I'd rather want to avoid running the MR with full-unit-tests. But if you think Frank that we should do full unit tests, then we'll do that...