In the aggregator code, WARN/ERROR takes precedence over STALE when computing the top_level_state. Was this an intentional design decision? If so, can I ask what was the reasoning behind it? I can understand ERROR taking precedence, but I am curious why WARN does as well. For example, a node that has stopped publishing entirely could be more critical than one that's still alive but reporting a warning, so I'm wondering what motivated this design.
In the aggregator code, WARN/ERROR takes precedence over STALE when computing the top_level_state. Was this an intentional design decision? If so, can I ask what was the reasoning behind it? I can understand ERROR taking precedence, but I am curious why WARN does as well. For example, a node that has stopped publishing entirely could be more critical than one that's still alive but reporting a warning, so I'm wondering what motivated this design.