A similar design to TCP/IP - sequence numbers, acknowledgements & resending capability on sender side
Sequence numbers on messages sent by producers coupled with some sort of persistence by the sender. This way when a gap is detected some sort of resend protocol can be implemented to achieve the delivery guarantee.
So the sender just keeps trying until it gets an ACK. Fair enough, though I imagine you'd want to run multiple nodes even with this.
Correct. Reliability and Availability are two separate but related goals.
This is a solved problem in most messaging protocols.