When querying the healpix pixels in a disc around the neutrino position, we have to make sure all pixels that fall within the rectangle are selected. That is the case if we take the longest side of the rectangle given by the coordinate uncertainties and devide it by $\sqrt{2}$ not $2$!
Probably some coverage estimates in the past were a bit too optimistic.
When querying the healpix pixels in a disc around the neutrino position, we have to make sure all pixels that fall within the rectangle are selected. That is the case if we take the longest side of the rectangle given by the coordinate uncertainties and devide it by $\sqrt{2}$ not $2$! Probably some coverage estimates in the past were a bit too optimistic.