After the !20258 (merged) got in, we managed to reproduce the crash at the reco level shown by the CI. Reproducing it allowed to understand it, and this MR fixes it.
In case we run the ITk reconstruction, the additional fields in the summery are used to decorate the track particle, otherwise they are skipped.
I have the tested both the ITk reconstruction and the Tier0 tests and everything is ok. I would expect the CI result to end successfully this time. If so, the MR can be merged in directly.