Fix issue with incomplete hashing in HLT Nav conversion
This MR fixes a very subtle bug in nav conversion. We were using a home made hashing functions to decide if two TEs were result of the same reco (i.e. can be merged). That hashing function resulted occasionally in the same hash value for actually differently made TEs. The code is now repalced by boost hash_combine.
In addition,
- revived standalone test (much faster than runnging derivation)
- setup unit tests
- added handy printer function in converter to aid future debugging
- improved messaging