perf(engine): include backpressure in newPayload latency metric

[mediocregopher]

The reth_newPayload latency timing previously only measured on_new_payload execution, starting the clock after backpressure and persistence waits had already completed. This hid real-world engine responsiveness issues from benchmark results — a change that caused 6x persistence backpressure could still show as a 7% improvement in all headline metrics.

Now latency = enqueued_at.elapsed() - explicit_persistence_wait, so backpressure (time the message spent queued because the engine was busy) is included in all headline metrics (Mean, P50, P90, P99, Wall Clock) while explicit persistence wait (the deliberate wait for previous block to persist) remains excluded.

Also adds Persist Wait as a row in the main benchmark comparison table with significance testing, so regressions in persistence throughput are visible at a glance instead of buried in a collapsed section.

Prompted by: mediocregopher

[mediocregopher]

derek bench blocks=10 warmup=0 abba=false

[decofe]

cc @mediocregopher

❌ Benchmark failed while running baseline benchmark (1/2). View logs

[mediocregopher]

derek bench blocks=10 warmup=1 abba=false

[decofe]

cc @mediocregopher

✅ Benchmark complete! View job

Benchmark Results

Metric	main	mediocregopher/bench-latency-includes-backpressure	Change
Mean	78.13ms	77.51ms	-0.79% ⚪ (±1.58%)
StdDev	43.90ms	43.49ms
P50	62.70ms	58.78ms	-6.24% ✅ (±4.83%)
P90	179.62ms	177.67ms	-1.09% ⚪ (±1.16%)
P99	179.62ms	177.67ms	-1.09% ⚪ (±1.16%)
Mgas/s	526.82	532.36	+1.05% ⚪ (±2.58%)
Wall Clock	0.80s	0.79s	-0.77% ⚪ (±1.51%)
Persist Wait	0.03ms	0.03ms	-10.63% ⚪ (±15.07%)

10 blocks, 1 warmup

Wait Time Breakdown

Persistence Wait

Metric	main	mediocregopher/bench-latency-includes-backpressure
Mean	0.03ms	0.03ms
P50	0.03ms	0.03ms
P95	0.06ms	0.06ms

Trie Cache Update Wait

Metric	main	mediocregopher/bench-latency-includes-backpressure
Mean	0.00ms	0.20ms
P50	0.00ms	0.00ms
P95	0.00ms	1.58ms

Execution Cache Update Wait

Metric	main	mediocregopher/bench-latency-includes-backpressure
Mean	0.00ms	0.00ms
P50	0.00ms	0.00ms
P95	0.00ms	0.00ms

Grafana Dashboard

Charts

Latency, Throughput & Diff

Wait Time Breakdown

Gas vs Latency

Grafana Dashboard

View real-time metrics

[mediocregopher]

derek bench blocks=10 warmup=1 abba=false

[decofe]

cc @mediocregopher

✅ Benchmark complete! View job

Benchmark Results

Metric	main	mediocregopher/bench-latency-includes-backpressure	Change
Mean	81.50ms	77.57ms	-4.82% ⚪ (±5.32%)
StdDev	45.15ms	44.20ms
P50	75.18ms	63.52ms	-15.51% ✅ (±14.25%)
P90	186.14ms	178.28ms	-4.22% ✅ (±2.57%)
P99	186.14ms	178.28ms	-4.22% ✅ (±2.57%)
Mgas/s	511.30	537.54	+5.13% ⚪ (±7.82%)
Wall Clock	0.83s	0.79s	-4.75% ⚪ (±5.36%)
Persist Wait	0.03ms	0.03ms	-4.83% ⚪ (±13.04%)

10 blocks, 1 warmup

Wait Time Breakdown

Persistence Wait

Metric	main	mediocregopher/bench-latency-includes-backpressure
Mean	0.03ms	0.03ms
P50	0.03ms	0.03ms
P95	0.06ms	0.08ms

Trie Cache Update Wait

Metric	main	mediocregopher/bench-latency-includes-backpressure
Mean	0.13ms	0.06ms
P50	0.00ms	0.00ms
P95	1.29ms	0.61ms

Execution Cache Update Wait

Metric	main	mediocregopher/bench-latency-includes-backpressure
Mean	0.00ms	0.00ms
P50	0.00ms	0.00ms
P95	0.00ms	0.00ms

Grafana Dashboard

Charts

Latency, Throughput & Diff

Wait Time Breakdown

Gas vs Latency

Grafana Dashboard

View real-time metrics