It was noticed today by @tstreble that unfortunately having the HGTD switched on by default in the dictionary generation breaks reconstruction from existing RDOs (e.g. those from the phys val campaign). Indeed, I see differences in the Identifiers written to output when running RunUpgradeTests.py and toggling this on/off, and so while jobs complete successfully and the results look normal when starting from HITS (because Identifiers are then consistent throughout), we can still see the effect that causes problems when starting from RDOs (where they are not, and so no tracks are found).
I've therefore tied the generation of the HGTD part of the dictionary to the HGTD_onDetFlag so that things will remain consistent for the moment, and the differences will only come in once we try to actually include the HGTD in the geometry fully.
