Pre-allocate memory for visit_dependencies_and_verify

[wqfish]

From running the benchmark, it seems like there are lots of memory allocation
and deallocation coming from visit_dependencies_and_verify. Pre-allocate the
memory if possible.

The capacity is what I guessed from printing the lengths. If there's a better
way to compute the estimated capacity, let me know.

[github-actions]

✅ Forge suite compat success on 79e566c8e0cf86948f975f3b8bc95e9441fce6bf ==> 56fa4e8d8368ed5c11c5a878022efd666f0b2605

Compatibility test results for 79e566c8e0cf86948f975f3b8bc95e9441fce6bf ==> 56fa4e8d8368ed5c11c5a878022efd666f0b2605 (PR)
1. Check liveness of validators at old version: 79e566c8e0cf86948f975f3b8bc95e9441fce6bf
compatibility::simple-validator-upgrade::liveness-check : committed: 12209.53 txn/s, latency: 2813.00 ms, (p50: 3000 ms, p70: 3100, p90: 3400 ms, p99: 3900 ms), latency samples: 407000
2. Upgrading first Validator to new version: 56fa4e8d8368ed5c11c5a878022efd666f0b2605
compatibility::simple-validator-upgrade::single-validator-upgrade : committed: 6044.30 txn/s, latency: 5594.71 ms, (p50: 6100 ms, p70: 6200, p90: 6400 ms, p99: 6700 ms), latency samples: 208400
3. Upgrading rest of first batch to new version: 56fa4e8d8368ed5c11c5a878022efd666f0b2605
compatibility::simple-validator-upgrade::half-validator-upgrade : committed: 5933.85 txn/s, latency: 5698.45 ms, (p50: 6300 ms, p70: 6400, p90: 6500 ms, p99: 6700 ms), latency samples: 204420
4. upgrading second batch to new version: 56fa4e8d8368ed5c11c5a878022efd666f0b2605
compatibility::simple-validator-upgrade::rest-validator-upgrade : committed: 9799.94 txn/s, latency: 3305.81 ms, (p50: 3300 ms, p70: 3900, p90: 4500 ms, p99: 4900 ms), latency samples: 326920
5. check swarm health
Compatibility test for 79e566c8e0cf86948f975f3b8bc95e9441fce6bf ==> 56fa4e8d8368ed5c11c5a878022efd666f0b2605 passed
Test Ok

• Grafana dashboard
• Humio Logs
• Axiom Logs
• Validator CPU Profile
• Fullnode CPU Profile
• Test runner output
• Test run is land-blocking

[github-actions]

✅ Forge suite realistic_env_max_load success on 56fa4e8d8368ed5c11c5a878022efd666f0b2605

two traffics test: inner traffic : committed: 13591.92 txn/s, submitted: 13592.03 txn/s, expired: 0.11 txn/s, latency: 2771.60 ms, (p50: 2700 ms, p70: 2900, p90: 3000 ms, p99: 3600 ms), latency samples: 5063220
two traffics test : committed: 100.02 txn/s, latency: 731.48 ms, (p50: 700 ms, p70: 800, p90: 800 ms, p99: 1000 ms), latency samples: 1700
Latency breakdown for phase 0: ["MempoolToBlockCreation: max: 2.261, avg: 2.194", "ConsensusProposalToOrdered: max: 0.168, avg: 0.166", "ConsensusOrderedToCommit: max: 0.064, avg: 0.055", "ConsensusProposalToCommit: max: 0.230, avg: 0.221"]
Max non-epoch-change gap was: 0 rounds at version 0 (avg 0.00) [limit 4], 0.48s no progress at version 5243099 (avg 0.07s) [limit 15].
Max epoch-change gap was: 0 rounds at version 0 (avg 0.00) [limit 4], 0.41s no progress at version 2101572 (avg 0.41s) [limit 16].
Test Ok

• Grafana dashboard
• Humio Logs
• Axiom Logs
• Validator CPU Profile
• Fullnode CPU Profile
• Test runner output
• Test run is land-blocking