Does batching help? Has somebody run the numbers? Doesn't the single transaction size scale linearly with the number of the inputs and outputs?
Can you show me to a transaction in the mempool with batched channel creation? (Even a test one)