generalBernoulli has true/false flipped

[dschulman]

In Data.Random.Distribution.Bernoulli, the documentation says the arguments are:
"generalBernoulli t f p" where value t is chosen with probability p and f with probability 1-p.

However, in the implementation, the order is reversed (f before t).
generalBernoulliCDF has the same issue.

Not sure if you'd prefer to update the documentation or the implementation.