One way to speed up the tree computation is to parallelize it. Currently, this is not possible because a single shadow project is used to compute the effect of each change. Instead, separate shadow projects should be used to compute the effects of changes in parallel. This will enable us to compute the change effects in parallel.