Hoi Daniele,

in principle  muon ID  already contains  a veto on  kaon ID  (that was
built in  to remove the kaon  in-flight decays some time  ago), so the
fact that you  see a small effect is maybe by  having a different veto
selector than what we use for kaon ID.

I think the  fix should be where the leading  lepton is identified. It
is a "feature" that kaon ID overrides muon ID. 

I don't think  it affects our signal. Muons that  are kaons should not
enter the signal. Of course, it does affect the depleted sample.