add data prescale

add 1% prescale to signal data as blinding technique.

We've been applying a prescale offline so far, but it would be better to include it at the AP level. Note that it doesn't affect the event count, so if efficiencies are calculated with that value as the denominator, its effect is automatically incorporated.

The prescale is implemented using pseudo-random numbers from the python random module with the seed set for reproducibility. When using ProdConf (as with AnalysisProductions), this seed is then incremented based on the input file names, which should mean that every run gets a unique but reproducible seed. No such incrementation is applied when running outside of ProdConf.