#11641 naively allocates the entire private log size to the ciphertext, but this is incorrect: there's also the tag, headers, encryption padding, etc. We need to account for this to produce a better limit.
Given users can introduce alternative encryption schemes, we could initially either choose to a) assume they're using our own, or b) set some upper bound for the overhead and hope their scheme does not require more metadata than what we allocate.
#11641 naively allocates the entire private log size to the ciphertext, but this is incorrect: there's also the tag, headers, encryption padding, etc. We need to account for this to produce a better limit.
Given users can introduce alternative encryption schemes, we could initially either choose to a) assume they're using our own, or b) set some upper bound for the overhead and hope their scheme does not require more metadata than what we allocate.