Use the real Browser Debugger SDK in the runtime benchmark

[watson]

What and why?

The Live Debugger runtime benchmark measured dormant overhead against a hand-written stub of the SDK's instrumentation runtime ($dd_probes/$dd_entry/$dd_return/$dd_throw), which could silently drift from what actually ships. This PR replaces that stub with the real published Browser Debugger SDK, loaded but kept dormant, so the measured overhead reflects production. It also records per-run provenance — the SDK version and the CDN build fingerprint — so any benchmark result can be tied to the exact build it was measured against.

How?

• Fetch the published datadog-debugger.js bundle from the CDN, inline it before the fixture runs, mock the probe-delivery poll same-origin to return an empty probe set, and call DD_DEBUGGER.init() after load so the SDK wires the real $dd_probes lookup while staying dormant (no active probes).
• Assert the SDK loaded and is dormant, and read its version from DD_DEBUGGER.version, failing the run if it is missing.
• Capture the CDN object's build fingerprint from the bundle-fetch response headers — publish date (Last-Modified) and S3 ETag (content hash, normalized from CloudFront's weak W/"…" form) — and assert both are present, since the /v7/ bundle reuses the same release-format version across many master builds and is therefore ambiguous on its own.
• Add an optional onResponse hook to @dd/core's doRequest so callers can read response headers (it previously returned only the parsed body); the benchmark uses it to read Last-Modified and ETag.
• Surface the version and fingerprint in the console summary, the PR comment (the hash labeled "S3 ETag" so it is not mistaken for a git commit SHA), and the results JSON, with unit tests covering the label formatting and weak-ETag normalization; update the live-debugger CONTRIBUTING.md.

[watson]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• Use the real Browser Debugger SDK in the runtime benchmark #440 👈 (View in Graphite)
• Add Live Debugger runtime benchmark coverage #429 : 1 other dependent PR (#438 )
• master

This stack of pull requests is managed by Graphite. Learn more about stacking.

[datadog-datadog-prod-us1-2]

🎉 All green!

🧪 All tests passed
❄️ No new flaky tests detected

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 4e565bd | Docs | Datadog PR Page | Give us feedback!}

[github-actions]

Live Debugger Runtime Benchmark

SDK-loaded dormant-probe runtime overhead, measured against an uninstrumented bundle in the same browser session.

Browser Debugger SDK: 7.4.0 · built 2026-06-23 · S3 ETag 0766e1c8

Browser	Workload	Quality	Per-call overhead upper
chrome	Hot	clean	<= 4.84 ns
chrome	Tiny	clean	<= 0.99 ns
firefox	Hot	clean	<= 9.89 ns
firefox	Tiny	clean	<= 0.00 ns
safari	Hot	clean	<= 15.39 ns
safari	Tiny	clean	<= 18.88 ns

Full diagnostics

browser  workload  quality  per-call overhead upper  overhead upper           95% CI        A/A diag         block CI  acf(1)   baseline  instrumented                         samples
-------  --------  -------  -----------------------  --------------  ---------------  --------------  ---------------  ------  ---------  ------------  ------------------------------
chrome   Hot       clean                 <= 4.84 ns       <= 13.35%    4.80..4.83 ns  -0.00..0.03 ns    4.80..4.84 ns   -0.10  55.647 ms     63.055 ms   102 (trim 20%, outliers 6.9%)
chrome   Tiny      clean                 <= 0.99 ns       <= 15.92%    0.74..0.95 ns  -0.01..0.02 ns    0.71..0.99 ns    0.03  48.073 ms     55.725 ms   102 (trim 20%, outliers 7.8%)
firefox  Hot       clean                 <= 9.89 ns       <= 38.02%    9.80..9.87 ns  -0.02..0.03 ns    9.79..9.89 ns   -0.00  39.900 ms     54.990 ms   102 (trim 20%, outliers 4.9%)
firefox  Tiny      clean                 <= 0.00 ns        <= 0.02%   -0.01..0.00 ns  -0.00..0.01 ns   -0.02..0.00 ns    0.25  97.980 ms     97.980 ms  102 (trim 20%, outliers 17.6%)
safari   Hot       clean                <= 15.39 ns       <= 61.66%  14.75..15.18 ns  -0.01..0.01 ns  14.64..15.39 ns    0.13  38.300 ms     61.300 ms   102 (trim 20%, outliers 2.9%)
safari   Tiny      clean                <= 18.88 ns      <= 724.80%  18.45..18.78 ns  -0.00..0.00 ns  18.33..18.88 ns    0.45   6.240 ms     50.870 ms   102 (trim 20%, outliers 3.9%)

Raw samples are in the live-debugger-runtime-bench-results artifact.