Currently by default we only trace if error only. This makes it hard to diagnose the issues (need to tell the user to bump up trace level). On the other hand, we don't want to spam the trace with /deployments from Portal polling. The proposed solution is.
1. for /deployments, if TraceLevel.Info, we will only trace if not 304 (NotModified). For verbose, we will trace everything. Any other trace level, behave as it is today.
2. for other calls (we assume it is due to user explicit action) such as /deploy or /settings, we will always trace them if TraceLevel is not Off.