webchannel/benchmark: add 100 pair run

webchannel/benchmark/lens/2026-10-13_06-08_omnient-delay-sweep-100pairs_JST/artifacts.md

@@ -0,0 +1,22 @@
+# Omnient delay-sweep artifacts
+
+[Report](report.md) · [Connection reuse audit](connection-reuse-audit.md)
+
+Each delay retains every attempted OFF/ON trial. Raw CSV/JSON, logs, result bundles, histogram bins, paired differences, and selection decisions are linked below.
+
+| M1 delay (ms) | Artifacts | Transport audit | Histogram input and bins | Paired differences |
+|---:|---|---|---|---|
+| 0 | [Artifacts](delay-0ms/artifacts.md) | [Audit](delay-0ms/connection-reuse-audit.md) | [JSON](delay-0ms/histograms/histogram-data.json) | [CSV](delay-0ms/histograms/pair-differences.csv) |
+| 50 | [Artifacts](delay-50ms/artifacts.md) | [Audit](delay-50ms/connection-reuse-audit.md) | [JSON](delay-50ms/histograms/histogram-data.json) | [CSV](delay-50ms/histograms/pair-differences.csv) |
+| 100 | [Artifacts](delay-100ms/artifacts.md) | [Audit](delay-100ms/connection-reuse-audit.md) | [JSON](delay-100ms/histograms/histogram-data.json) | [CSV](delay-100ms/histograms/pair-differences.csv) |
+| 200 | [Artifacts](delay-200ms/artifacts.md) | [Audit](delay-200ms/connection-reuse-audit.md) | [JSON](delay-200ms/histograms/histogram-data.json) | [CSV](delay-200ms/histograms/pair-differences.csv) |
+
+## Infrastructure interruptions
+
+- [delay-100ms-pair-51-locked](infrastructure-attempts/delay-100ms-pair-51-locked/README.md)
+- [delay-200ms-pair-76-locked](infrastructure-attempts/delay-200ms-pair-76-locked/README.md)
+- [delay-50ms-pair-26-locked](infrastructure-attempts/delay-50ms-pair-26-locked/README.md)
+
+Histogram PNG and SVG files are generated with matplotlib. Each metric shares 10 ms bin edges across all delay settings. Exact bin counts, renderer version, generator SHA-256, and input SHA-256 manifests are saved with each delay's evidence.
+
+[Build log](logs/build.log) · [Successful trust preflight](logs/trust-preflight.log). Setup and preflight trials are not included in the measurement pairs.

webchannel/benchmark/lens/2026-10-13_06-08_omnient-delay-sweep-100pairs_JST/connection-reuse-audit.md

@@ -0,0 +1,18 @@
+# Omnient delay-sweep connection reuse audit
+
+[Report](report.md) · [Artifacts](artifacts.md)
+
+Per-delay audits retain every trial's HTTP/2 connection evidence, initial early POST status, and M1/M3 dispatch. HTTP 400 confirms rejection; client evidence does not identify the server-side cause.
+
+- [0 ms connection reuse audit](delay-0ms/connection-reuse-audit.md)
+- [50 ms connection reuse audit](delay-50ms/connection-reuse-audit.md)
+- [100 ms connection reuse audit](delay-100ms/connection-reuse-audit.md)
+- [200 ms connection reuse audit](delay-200ms/connection-reuse-audit.md)
+
+## Infrastructure interruptions
+
+Install/launch failures produced no paired benchmark measurement. Their files are retained separately; completed measurement pairs excluded by the histogram rules were not repeated. The interrupted pair was restarted as adjacent OFF/ON trials after reconnecting.
+
+- [delay-100ms-pair-51-locked](infrastructure-attempts/delay-100ms-pair-51-locked/README.md)
+- [delay-200ms-pair-76-locked](infrastructure-attempts/delay-200ms-pair-76-locked/README.md)
+- [delay-50ms-pair-26-locked](infrastructure-attempts/delay-50ms-pair-26-locked/README.md)