Need to track NAT exhaustion

One of the ways a flow can fail is due to resource exhaustion. The best current example is NAT exhaustion, e.g., running out of ports in a NAT Pool for making outbound connections. Here is where we handle the generic failure of a layer to generate an action descriptor. We need to hook in some basic statistics about things like NAT exhaustion. That could be directly at the above location, or possibly here which is where the Port object itself would get an error related to resource exhaustion from processing an inbound packet (either using a cached UFT entry or processing each layer serially).

Start by reading the generic layer failure handling in opte/src/engine/layer.rs near line 949 and the Port error path in opte/src/engine/port.rs near line 1653. Trace how resource-exhaustion errors from inbound packet processing reach those locations, then determine where basic NAT-exhaustion statistics should be recorded. Done means NAT Pool port exhaustion is represented in the requested statistics.