Track fakes update

New fake systematic variation that doesn't rely on truth information, for use in MC16 where truth strategy results in incorrect truth match probabilities for standard samples.

See here for motivation.

Opening an MR while final validation of implementation in CP tool vs Ntuple-based version is ongoing to allow people to comment on implementation.

FYI @dshope @npetters @mdanning @cpollard @mkagan @bnachman

Update: From validation plots here, new function seems to be doing what it should.