LEVM: Separate get_state_transitions from execute_block

[JulianVentura]

Currently, the execute_block function returns the BlockExecutionResult with the account updates, which are calculated by making a call to get_state_transition function. This is ok when we want to execute just one block, but if we want to execute blocks in batches we don't really need to get the account updates after each block execution, we could just obtain it at the end instead.

We want to separate the block execution from the accounts update calculation.

[JereSalo]

Yeah, as Juli says. The idea of this is to have execute_block without get_state_transitions. This would mean that in every place that we call execute_block right now we have to do get_state_transitions manually or wrap it into some other function if that's considered appropriate.
The advantage of this is that we won't be getting the account updates in cases in which we don't want to commit to the actual database! And one example of that is what he mentioned, executing blocks in batches and committing only after X blocks.

[JereSalo]

A follow up for this PR could be implementing add_blocks_in_batch for LEVM or something like that.