Exclude could bring down whole cluster

In some scenarios when using a small cluster and the processes have a suboptimal layout doing an exclude can bring down the whole cluster (e.g. when the cluster has not enough log processes to be recruited). The only way out of this is to add additional resources as the include command will not work. It would be great to 1. make sure that the include command can be used in such scenarios and 2. if the exclude command could check if enough processes are available to recruit the transaction system before doing the exclusion.

Start by locating the include and exclude command handling and the transaction-system recruitment checks; reproduce the small-cluster scenario described in the issue. Done means include remains usable after a risky exclusion and exclude refuses to proceed when too few processes are available.

Written by the indexing model from the issue text.