Cleans up workspaces from this job's old builds on other slave nodes.

By default, a slave node will leave a build's workspace on the slave node's filesystem indefinitely (until it gets overwritten by a new build). When there's a (large) pool of persistent slave nodes available to run a (large) number of builds, eventually every slave node ends up with a copy of every builds' workspace. This is a waste of hardware resources as Jenkins will only look at the workspace on the most recently used slave node.
This functionality triggers a clean up of the workspace on all slave nodes that are not currently being used so that, on average, the pool of slaves will only contain one workspace for each job.