feat: optimize phase0 attestation processing

[EchoAlice]

Optimize Phase0 Attestation Processing

Summary

Optimizes epoch_processing::get_inclusion_delay_deltas() by restructuring attestation processing flow, reducing epoch boundary block processing time by 88.8%.

Problem

The original implementation iterates through all pending attestations for each validator, creating a significant performance bottleneck in state_transition().

Solution

Restructures attestation processing to perform a single pass through pending attestations:

// Before
for validator in unslashed_validators {
    for attestation in pending_attestations {
        // Process each attestation for every validator
    }
}

// After
for attestation in pending_attestations {
    let validator_indices = calculate_validator_indices(attestation);
    // Process attestations once, mapping delays to validator indices
}

Performance Impact

Tested at slot 3,200 (20,488 validators, 225 pending attestations):

Processing time: 1.81hrs → 12.2 minutes
8.93x faster

Testing

Verified correctness against reference implementation
Benchmarked with mainnet data at slot 3,200

Next steps:

• Similar optimizations planned for Altair, Bellatrix, Capella, and Deneb forks