Waku v2 store node retention boundaries and offline delivery guarantees

Waku v2 routes messages by content topic rather than recipient address, and store-and-forward nodes can retain messages for offline peers. However, retention is operator-configured rather than a protocol-wide guarantee, and delivery is best-effort. When no store node is available while a recipient is offline, a message may be dropped without retry.

The unresolved behavior concerns how applications should model delivery guarantees when store node availability and retention policies vary across operators. If a client publishes to a content topic while the intended recipient is disconnected, and the only available store nodes have short retention windows, the protocol does not define a fallback or retry mechanism. This creates uncertainty about whether an application can treat store-and-forward as a reliable mailbox or must implement its own acknowledgment layer.

What are the documented boundaries of Waku v2 store node retention relative to message expiry? How should a client application interpret delivery semantics when a store node drops a message due to operator policy before the recipient reconnects? Does the protocol specify any behavior when no store node is reachable during an offline recipient's absence?