Most of the details of NAT traversal (for GIMPS-aware NATs) are a local
implementation matter. However, it might be useful to document how NAT traversal
is actually possible, in order to validate that the protocol itself is not
NAT-unfriendly, and also to provide some background rationale for some of the
details of the fundamental specification. The processing details could go in an
informative annex to the main GIMPS specification.