bug in 1d and multivariate

removed condition that seemed to create very large support
in sum hills calculation and resulting into a very large computational
time. Credits to Carolyn Phillips from the mailing list.