Block processing optimization

[chetanyb]

Summary

Adds comprehensive metrics instrumentation and justifications caching to identify and optimize block processing bottlenecks. Addresses #444

Metrics Added:

• Granular block processing breakdown (header, attestations, operations)
• State transition internal metrics (justification build, state update)
• Block processing path counters (cached vs computed state)
• Cache performance tracking

Caching Implementation:

• Justifications HashMap caching between parent/child blocks
• Configurable via cache_justifications flag
• Integrated into state transition flow

Key Findings

Cache currently degrades performance - needs redesign before merging:

• Current performance: 93.3 μs (with cache at 1.3k slots) vs 21 μs (without cache at 300 slots)
• Root cause: HashMap cloning overhead exceeds build cost
• TODO:: Add cache eviction (grows unbounded)
• TODO: Move from cloning entire HashMap on every lookup

Testing

Tested on AWS deployment at 1.37k slots with lean-quickstart