Investigate addition of socket safeguard

Each endpoint maintains infrastructure to make sure that the underlying channel is cleaned up if needed, but it would be helpful as an additional safety-net to keep a central component that tracks the number of channels on a per-instance basis and reports (and maybe even cleans up) channels that are hanging around even if they shouldn't.

Right now there is no evidence that such a problem exists, but as a future enhancement safety net it can make sense to investigate.