BlockSynchronizationView/Event & simple barrier (block metadata txn)

[gelash]

This should be an improvement over the current state of affairs, as finishing of block metadata txn would trigger everything that has done a read already (timestamp, randomness) to re-execute. With this change, workers will wait to start, which the blockmetadata transaction will allow after it finishes. saves wasted concurrent work and invalidation, but to discuss:

• should we do a condvar so workers are not busy waiting?
• should we pre-populate/hint the entries (timestamp, randomness), and allow the transactions to start earlier? Should be consistent with this design too though as we could just move the signal_can_start on write barrier earlier (but maybe we want to do this differently?)

Type of Change

• [x] New feature
• [ ] Bug fix
• [ ] Breaking change
• [x] Performance improvement
• [x] Refactoring
• [ ] Dependency update
• [ ] Documentation update
• [x] Tests

Which Components or Systems Does This Change Impact?

• [x] Validator Node
• [ ] Full Node (API, Indexer, etc.)
• [ ] Move/Aptos Virtual Machine
• [ ] Aptos Framework
• [ ] Aptos CLI/SDK
• [ ] Developer Infrastructure
• [ ] Other (specify)

How Has This Been Tested?

Testing two types of barriers in proptests