WsClient calls should be aware of maxreqentity behavior

in certain cases WsWorkunitsStub::wUCreateAndUpdate creates the workunit but fails to populate it.  Since the cluster happily creates the empty workunit in unknown state and the api fails to populate it the API should clean up after it self.  It would be helpful and easier to debug/trace/diagnose issues if in the catch the code would try to update the workunit and set the error the API logged into the workunit and mark the workunit failed. 

Specifically the case the ECL was so large it was exceeding the maxRequestEntityLength value of the cluster.  However the only thing that the API saw was simply a "broken pipe" and it left the workunit it created in unknown state.