We're curious if when we load that same "Person" object into Context in Aware if a data set is "loaded" of the equivalent huge size of the XML or JSON we observe in REST.
No, it's not equivalent. Aware IM will only load nested references if you access them by rules. Otherwise, it's very well optimised. With JSON and XML Aware IM HAS TO produce the entire tree, but it doesn't have to do it when an object is just loaded into Context.
If this is the case is it wise to avoid too many levels of nested objects when you are designing your AwareIM data model to reduce Context "Data Volume"??
No, you shouldn't worry about the level of nesting.