feat: share sender recovery cache across tx processing paths

[shekhirin]

Summary

• Share a bounded sender recovery cache across transaction pool ingress, network recovery, and payload prewarming so recovered senders can be reused instead of recomputed.
• Thread the cache through node, network, and EVM wiring, and use it when converting pooled transactions and execution payload transactions.
• Add coverage for cache population and failed recovery behavior.

[Rjected]

derek bench

[decofe]

cc @Rjected

✅ Benchmark complete! View job

Benchmark Results

Configuration

• Derek command: derek bench blocks=500 warmup=125 baseline=main feature=alexey/shared-sender-recovery-cache bal=false cores=0 run-pairs=6 otlp=true metrics=false slack=always

⚠️ Feature is 29 commits behind main. Consider rebasing for accurate results.

Metric	main	alexey/shared-sender-recovery-cache	Change
Mean	30.06ms	30.07ms	+0.01% ⚪ (±0.75%, floor 1.20%)
P50	23.84ms	23.86ms	+0.06% ⚪ (±1.07%, floor 1.20%)
P90	48.86ms	48.89ms	+0.07% ⚪ (±1.93%, floor 1.35%)
P99	164.13ms	156.67ms	-4.55% ⚪ (±3.66%, floor 5.00%)
Mgas/s	1167.03	1161.87	-0.44% ⚪ (±0.55%, floor 1.20%)
Wall Clock	90.73s	90.69s	-0.05% ⚪ (±0.72%, floor 0.70%)
Persist Wait	29.19ms	29.54ms	+1.21% ⚪ (±1.62%, floor 5.00%)

500 blocks, 125 warmup, 6 run pairs

Wait Time Breakdown

Persistence Wait

Metric	main	alexey/shared-sender-recovery-cache
Mean	29.19ms	29.54ms
P50	0.02ms	0.02ms
P95	271.47ms	268.46ms

Trie Cache Update Wait

Metric	main	alexey/shared-sender-recovery-cache
Mean	0.86ms	0.87ms
P50	0.00ms	0.00ms
P95	7.12ms	7.07ms

Execution Cache Update Wait

Metric	main	alexey/shared-sender-recovery-cache
Mean	0.00ms	0.00ms
P50	0.00ms	0.00ms
P95	0.00ms	0.00ms

Observability

• Logs
• Traces

Charts

Latency, Throughput & Diff

Wait Time Breakdown

Gas vs Latency

[pepyakin]

derek bench mode=rpc

[decofe]

cc @pepyakin

❌ Benchmark failed while parsing results. View logs