I'm curious: when you have a deeply-nested trace of spans (Mastra Workflows, in my case) is there any deduplication done on the inputs and outputs? If you have not-small context in or results out (10s of kb) and they are identical "on the way in" and/or "on the way out", is there anything done to dedup all that data? Also, since Postgres is the persistence layer, are TOASTing our rows when the amount of data is >8kb?
