Not so “universal” if it’s per-observer, right?

There's a universal partial order of events. If something's in your future light cone, or your past light cone, that fact is independent of the observer. For the things that are not dependent on your present, and do not affect your present, different observers can have different understandings of the order.

Statements “something is in your future” and “that fact is independent of the observer” seem incompatible. When you are saying “your”, you are talking about that particular observer, so you are negating the “independent” part.

It’s dependent on a reference location not the existence of an observer at that location. Further you can compute these timelines for that location without physically being at that location.

Thus given location X you can get a universally agreed upon ordering. Or alternatively you can devise locations from various sets of orderings. A related example might be using multiple audio recordings to determine when and where various shots were fired.

Reference location in where? In multiverse? But then again there is a dependency on having a correlation with something that can contain bits of information.

An example: You on Earth see events X and Y happening at Alpha Centauri and Proxima Centauri respectively. If the observers at Proxima Centauri didn't see event X happening at Alpha Centauri before Y happened there, and the observers at Alpha Centauri didn't see Y happening at Proxima Centauri before X happened there. Then the events can be seen in either order by an external observer depending on where they are and how fast they move. If an observer at Alpha Centauri or Proxima Centauri did see the other event before their event happened, then they have a well defined order ('is inside the light cone').

Which isn’t particularly meaningful when each location can agree on the ordering observed at the other locations.