Incorrect transverse mass `mt2` definition

[Jerry Ling]

LorentzVectorHEP.jl/src/cartesian.jl

Lines 4 to 5 in 189aa11

 mt2(lv::LorentzVector) = lv.t^2 - lv.z^2 
 mt(lv::LorentzVector) = mt2(lv)<0 ? -sqrt(-mt2(lv)) : sqrt(mt2(lv)) 

idk why we didn't do the naive px^2 + py^2, because that seems to be correct

[Jerry Ling]

here our mt^2 is effectively

E^2 - pz^2 = m^2 + px^2 + py^2 + pz^2 - pz^2

which only gives me correct plot if particle is massless

[Jerry Ling]

but it's how ROOT does it https://root.cern.ch/doc/master/TLorentzVector_8h_source.html#l00513

@aminnj halp

[Jerry Ling]

I think we're missing a 2-particle version of mt2 and mt because that's apparently different from mt(vec1 + vec2), makes sense