Yeah, it's not going to be a big storage hit in reality (and even if it were, you could just rely on file system level compression/deduplication). I, like the author, just enjoy musing about the "perfect" way to organize a music library.
Deduplication should be done by algorithm that detects that it is same binary copy -- not through tags.