I'm going to suggest that we abandon the ResultSetAdapter concept. There's not enough "meat" to justify including it in the platform. By the time I was done stubbing it out last night, the class was mostly done. My conclusion is that this can an be more efficiently done at the application level. It gives more control to the app developer and eliminates the need to make modifications to JSONSerializer and CSVSerializer to accommodate it.
Unless there are objections, I'm going to remove this class from SVN.
