Fix EtaPtFilter to avoid some missing parent links for leptons

For ggF PowHeg/Pythia H4l samples of mc16/mc20, there is a rare occasion when the EtaPtFilter of the truth (HepMC to TruthEvent) removes the link between a status 23 lepton (el/mu) and its status 1 child. This limits the navigation to the status 1 lepton. (These are leptons with eta > 5.5, but we would still like to keep track of them.) The solution, agreed with @zmarshal, is to allow all lepton pass through, and this solves the problem. Note that the EtaPtFilter reduces the truth size by a bit less than 10% on the AOD, and so is useful for this.

To get this to work in the nightly, we uncommented the code loading McAodBuilder, as suggested by @jchapman. This apparently was commented out during the CA migration, and should be in the standard production. This runs fine for my tests... We can see for all tests. This might change the AOD size for the continuous testing, since I believe it is adding back in the filtering which was missing for a while. I see 4% file size reduction for ggF H4l PowHeg sample.

I also made CA config files for McHepEvent and xAODTruth readers and tested then running them 'by hand' on HITS and AOD files, agreed with @jcatmore. This was used in my tests.