bind failures while running tests
1) Fixed bunch of bind failures where the server is not stopped after
the test causing the bind failures.
2) Added instrumentation to report the port for http when the bind
failure happens, so that we can dig through the log to find the last
started service.
3) Fixed the resource loader to throw error when resource is missing
instead of failing with NPE.
4) Fixed Streaming Client to throw the actual error instead of failing
with NPE on bind failures.
5) Increased the sleep time for Zone Rebalance test as 2 ms is not
enough on my machine. 5 ms sort of fixes the problem, but only time will
tell