Atomic Block Production

[rpanic]

Currently, block production isn't atomic and consistent when it comes to DB writes. If the block production fails in the middle of production or between production and result generation, inconsistent states can occur.
To fix this, two things should be implemented:

• Make result generation a seperate step in the production pipeline and enable it resume operation in case the sequencer crashed after production and before result generation
• Create a DB transaction pattern that allows us to encapsulate all of the DB writes done during block production in a single transaction