v1.1.0

• Improved relay throughput by reducing allocations across host, client, receive, and broadcast hot paths.
• Added relay backpressure handling to keep slow or overloaded recipients from degrading room performance.
• Made unreliable relay traffic more robust by dropping payloads that are too large or when a recipient's reliable queue is already backed up.
• Added protection for reliable relay queues by disconnecting recipients whose reliable backlog grows beyond the configured threshold.
• Optimised single-recipient and broadcast relay paths, including special handling for one-client broadcasts.
• Reworked relay frame creation to use LiteNetLib pooled packets where possible, stack allocation for small frames, and shared array pooling for larger frames.
• Avoided unnecessary payload copies by reading incoming packets as spans and writing relay frames directly into the target buffers.
• Reduced allocation pressure when relaying client-to-host and host-to-client data.
• Introduced a reusable relay backpressure policy with clear thresholds for dropping unreliable traffic and disconnecting reliable recipients.
• Added safeguards for oversized unreliable relay payloads before attempting to send them.
• Added handling for LiteNetLib TooBigPacketException so rejected relay payloads are logged cleanly instead of flowing through the generic send failure path.
• Snapshot broadcast recipients before sending, reducing lock contention while preserving consistent recipient handling.
• Added log messages for dropped oversized unreliable payloads.
• Added log messages for unreliable payloads dropped due to reliable queue backpressure.
• Added log messages when recipients are disconnected because their reliable relay queue is backed up.
• Added log messages when LiteNetLib rejects a relay payload as too large.

Full Changelog: v1.0.0...v1.1.0