Better L2 block representation

[eyusufatik]

We lack two things:

1. No state root in the block header, so the hash of the block does not cover storage root.

   Right now the sequencer does not actually sign the state root, which might be problematic later on. We must include the state root in the header.

2. No header / header + txs seperation → No sov-tx merkle root

   Having small data that tells somethings about the chain is good.
   Every chain that I'm aware of has

   Block = header + transactions

   That can become useful in many contexts.

With these changes the proposed header is as follows:

struct L2Header {
  l2_height: u64,
  prev_hash: [u8; 32],
  state_root: [u8; 32],
  // these two maybe removed or not
  // depending on system tx decision
  // da_slot_height: u64,
  // da_slot_hash: [u8; 32],
  // da_slot_txs_commitment: [u8; 32],
  l1_fee_rate: u128,
  tx_merkle_root: [u8; 32],
  // maybe removed due to 
  // deposit_data: Vec<Vec<u8>>,
  timestamp: u64,
}

struct SignedL2Header {
  header: SoftConfirmationHeader,
  hash: [u8; 32],
  signature: Vec<u8>,
}

struct L2Block {
  header: SignedSoftConfirmationHeader,
  txs: Vec<SovTx>,
}