cdf output ids are inconsistent at upper end

[lmullany]

for a very small number of cdfs (n=736), the upper end of the cdf output_type_id is inconsisent. There are only two targets (ili perc and season peak perc) ,and so the upper end of the cdf should be 100.0, but is labelled NA in some cases. We can manualy correct this, and doing so is important for issue #28