Implement a LedgerBackend for benchmarking ingestion performance as a function of ledger size

[tamirms]

Given the following configuration options:

• minimum number of transactions / operations per ledger
• ledger close time
• queue capacity

The LedgerBackend will start prefetching ledgers from the data lake and, using #5479 , it will combine consecutive ledgers until we have a ledger which meets the minimum number of transactions / operations per ledger. Then the resulting "fat" ledger will be emitted at a rate matching the configured ledger close time (we can default to 5 seconds per ledger). If the fat ledgers are emitted at a rate faster than they can be consumed, the LedgerBackend will push the fat ledgers onto a queue. Once the queue has exceeded its capacity we can terminate with an error because we know that the consumer of the LedgerBackend is not able to keep up with the current throughput settings.