Slim the LeptonTagger for r21.2
In some very rare cases, there will be a lot of tracks (i.e. 55 tracks) within cone size delta R < 0.3 regards to the lepton direction.
It will create too many combinations during the vertex fitting procedure. And will take a lot of memory. We slim the LeptonTagger algorithm by adding a limit on the number of tracks we used for EXTRA vertex fitting.
The effect of this change to the PromptLeptonImprovedVeto outputs should be negligible.
Edited by Atlas Nightlybuild