For simulations. These methods, given an elec.data object, make a
``truth''---i.e. a different vote count---that meets the same precinct and
tot.votes structure, but has potentially different results and outcomes.
make.truth.opt.bad
makes the ``optimally worse truth'', where the
error needed to flip the winner and runner-up is packed into as a few
precincts as possible.
make.ok.truth
makes the truth have the same outcome as the reported,
but some errors here and there.
Warning: if bound is WPM this error is made by simply adding the max amount of error
to the first loser's total (so that total votes may in this case exceed the total votes
of the precinct)--this could potentially cause trouble. Be careful!
make bad truth as described in Stark's paper (assuming fixed precinct size)