A *regular event* is defined as an event with a kind `1000 <= n < 10000`.
Upon a regular event being received, the relay SHOULD send it to all clients with a matching filter, and SHOULD store it. New events of the same kind do not affect previous events in any way.
Upon a replaceable event with a newer timestamp than the currently known latest replaceable event with the same kind and author being received, the old event SHOULD be discarded,
effectively replacing what gets returned when querying for
Clients SHOULD use the `supported_nips` field to learn if a relay supports this NIP. Clients SHOULD NOT send ephemeral events to relays that do not support this NIP; they will most likely be persisted. Clients MAY send replaceable events to relays that may not support this NIP, and clients querying SHOULD be prepared for the relay to send multiple events and should use the latest one.