When Fusaka activates on L1, the batcher needs to behave differently.

It needs to send blob transactions with cell proofs rather than blob proofs. See this doc for more info. We should bring in code from upstream geth to support this, we may need to bake it in if it gets removed from upstream geth.

We probably need to coordinate activating the new behaviour with Fusaka activation (cell proofs unlikely to work before Fusaka).

For a v0, we should just add a CLI flag to control the type of transaction used. This allows us to successfully defend Fusaka by having chain operators restart their batcher close to Fusaka activation with the new behavior enabled.

We also need to consider EIP 7825,: what if calldata batches use too much gas? Also, the intrinsic costs change, which could affect the minimum.