According to that definition, if your specific time constraints are "between 10ms and 1000ms", wouldn't HTTP be sufficient?

Seems like your argument for if this is "realtime" or not depends on parameters not specific by either you or the project.