Per-benchmark timings against baselines and comparable crates
(wide, bevy_ecs, and others — see
docs/benching-standards.md).
wide crate.<suite>/<suite>-bench-results.json
per suite, same convention as the Tests page's
<crate>-test-results.json. None of the existing
bench-vs-*.yml workflows emit that shape yet — they currently
write an HTML report directly (see docs/benching-standards.md).
A card here reads "No results yet" until its own bench workflow is updated
to also write that JSON alongside its existing HTML report, and
Deploy Site is run afterward to publish it. Wiring one
workflow at a time is deliberate — same trigger-based discipline as
everything else in this project.