We are not pre computing all node status addequately when creating and
updateing individual node status. This allows us to quickly traverse the
tree when scheduling without looking at and incoming edges for checking
blocked builds.