deadzones not properly handled when consecutive positions reside in different intervals

[andrewdavidsmith]

I've had problems with the bin scaling factor going negative if some deadzones are consecutive, but not merged. This should not be provided as input, but the algorithm should be able to handle non-overlapping intervals even if they should have been merged due to covering adjacent positions. The cause in my case was incorporating unassembled regions without merging intervals. This should be fixed in both the deadzones program and in rseg, specifically in how the deadzones are loaded for adjusting the scaling factors.