The recursive aggregation circuit currently has ~101K rows for the Public table.
This could basically entirely removed (down to a single digest), if we were to hash the entire set of values and checking the hash inside the circuit.
Because we have 101K values currently, that would incur a ~12.7K increase in Poseidon2 rows (note that if #380 was implemented first, we'd trim away a large portion of that so the overhead on Poseidon2 would be lighter).
The recursive aggregation circuit currently has ~101K rows for the Public table.
This could basically entirely removed (down to a single digest), if we were to hash the entire set of values and checking the hash inside the circuit.
Because we have 101K values currently, that would incur a ~12.7K increase in Poseidon2 rows (note that if #380 was implemented first, we'd trim away a large portion of that so the overhead on Poseidon2 would be lighter).