Improve support for many-to-many topologies. #112

Hello, I've been working on making NORM be more tolerant of many-to-many topologies. To that end I've drafted four mostly independent prs.

• Add sender inactivity timeout auto purge #108: Sender inactivity timeout auto-purge and object table crash fix
• Add node stats telemetry and fair rx limits #109: Node stats telemetry API, per-sender RX cache limits, and fair round-robin event queues
• Add stream buffer full backpressure API #110: Stream buffer full backpressure notification API
• Add adaptive CC probe scaling and NACK limits #111: Adaptive CC probe scaling, piggybacking, and NACK failsafe limits

Everything is designed to be strictly opt-in and backwards compatible with existing deployments. I am happy to tweak the interfaces, naming conventions, or implementation details to better align with the project as needed.

Thanks!