Don't log full stack trace on Elasticsearch timeouts

Observed when Usergrid application servers were running out of memory, thousands of Elasticsearch exception stack traces were logged. Change this to log only the exception message and no need to print the full stack trace.