I voted November, which is the peak estimate but I do think there is a non zero probability it won't launch at all.
SpaceX need to start launching their backlog of manifested launches, continuing prep for Commercial Crew (don't they have a LES test of some sort scheduled for 2015?) and get an initial launch of FH.
This suggests that while doing this they will also be training up staff to raise numbers, although I think automation can cover a lot of work.
I think the FH launch (and just as importantly next
launch) will depend on exactly how close to 3 F9 cores is
the 1st stage (and wheather propellant transfer will be demonstrated on the first flight)?
Best case. All
F9 1st stages can be part of an F9 or FH as they have the load carrying fittings and propellant transfer lines fitted as standard. IE Any first stage can be an F9, FH core or FH 1st stage booster.
Worst case. F9 1st stages, FH cores and FH boosters are all substantially different, especially in the nose end, with either no, from/to or to/from 2 part propellant couplings, along with heavier unique structural designs to carry the different loads.
Both will superficially look the same but the latter will need a lot
more structural and stress analysis, not just of 3 different structures but of their combined
option in this is to let the schedule slip, but I hope they won't do this too much.
BTW The more those stages are like an F9 the more confidence you can have in a Bayesian analysis of the probability for launch success. If they are very similar you got a pretty good chance of a 1st liftoff, otherwise standard statistics say 50/50 of an OK launch,