[Rivet] PartonicTops with em shower

[Markus Seidel]

Dear Rivet authors,

as you know PartonicTops projection is important for many CMS top 
analyses at 8 TeV :s
People stumbled upon the following problem: If there is a photon->l+l- 
within the shower, the top mother is flagged as leptonic.

Maybe we can replace
 > const bool prompt_e/mu = any(...);
by
 > const bool prompt_e/mu = (count(...) % 2);
and
 > if (prompt_e/mu && (...)) return true;
by
 > if (prompt_e/mu && !prompt_mu/e && (...)) return true;

For the hadronic part there needs to be an `xor` between `prompt_e` and 
`prompt_mu` at least...

If you prefer to not mess around with the logic, we may just require 
that the lepton does not have any photon mother...
 > ... && !p.hasAncestor(PID::PHOTON)

It seems to be hopeless to account for all possibilities of a full weak 
shower but at least the case of em shower is very common (at least in 
Pythia 8?).

Cheers,
Markus