The IDE (in this case atom) spins up the process and shuts it down when necessary, and then both sides communicate to each other using an implementation-defined IPC mechanism.
So it's a server more in the sense that the IDE requests info as a client, and then the server provides it (pull, not push).
If a program runs in a separate process and provides data/processing to other connected programs through some standard protocols, they are a server.