* Reduce the min size for large preimages in devnet to make it easier to reach the limit.
* Provide a buffer between target and max batch size to avoid the batcher splitting the large tx across multiple smaller frames.