Document consistent load and nodes not being cycled out edge case

We should document the edge case that exists with Escalator in that when there is a consistent amount of base load and Escalator always maintains a certain amount of nodes, it will never cycle out the oldest nodes first, as they'll always be used to run workloads.

No documentation file or test is named in the issue. Review the existing Escalator documentation and the node-cycling behavior, then document the consistent-base-load edge case and how maintaining a fixed node count affects which nodes are used; done means the behavior is clearly recorded for users.

Written by the indexing model from the issue text.