Thanks @phroi, this was honestly helpful; I carefully traced the findings and they apply: subset finalization makes omission targetable by lane, and pending_intent_count currently provides no effective close invariant. The distinction between one-shot authority and the source of that authority also clarifies the remaining uniqueness problem. I will look into the iCKB path separately.
I understand your time constraint and would respect that. I will review Deposit-Paired Voting publicly when I can give it proper attention. Thank you again for taking the time to examine the work.