It is hardly a strong argument. His numbers bounce around in a range that is normal for him. And the density of data are not enough to show that the increase in the latter half of the Vuelta is a statistically robust trend. Looking at all the data over time, the numbers bounce around quite a bit and there are no analytical error bars. Further, I'm not aware of any evidence that hemoglobin numbers generally diminish throughout a stage race. And one of the commenters on that page make a good point, that varying degrees of hydration/dehydration can influence those numbers.
I"m not a blind adherent to the proposition that Horner was clean - I definitely have my doubts, but those data don't make a persuasive case otherwise.