Negative "situations" in validphys reports

Created by: scarlehoff

Turns out that the "negative replicas" in the reports were due to the central value - standard error going negative, with no actual replica producing a negative result.

For instance, I did a quick DIS fit and I checked that no negative results were produced:

> from validphys.api import API
> from validphys.convolution import predictions
> inp = {'dataset_input': {'dataset' : 'POSXGL'}, 'theoryid' : 53, 'use_cuts':'nocuts', 'pdf':'091020-nnpdf40_dis'}
> preds = predictions(API.dataset(**inp), API.pdf(**inp))
> any([any(preds.get(i)<0) for i in range(1,51)])
False

However if I look at the plot the band goes negative: matched_positivity_from_dataspecs10_dataspecs0_plot_positivi.pdf

To avoid future unnecessary discussion I don't know whether it would be good to either add a disclaimer at the beginning or change the meaning of the error bar for the positivity reports.

Note: this also means that the positivity is also being checked up to the 10^-6 zone which is good enough I think.