Current implementation computes and seals a block immediately after seeing a single L2 tx. This is undesirable in a lot of cases when you want to test scenarios closer to production with multiple txs making it into the same block.

I propose we allow two modes of operation:

• Manual block sealing - there is an existing mining mechanism that we can reuse (currently always mines empty blocks which is of limited value anyway)
• Automatic block sealing every so often (configurable on startup)