Many of the jobs that produce artifacts or containers don't do cleanup when done, assuming that the node on which they were built will be a PSI temporary node and the whole machine will be garbage collected.

But because scm polling often fails due to the use of temporary nodes, we'd like to use more permanant nodes.

Using permanent nodes means that garbage collection is critical to ensuring the machines are healthy and don't run out of disk (eg., for theia builds).

So we need to add a step to clean up workspaces for successful builds... but NOT purge the scm poll log data, so that we can fix CRW-2542.