feat: Add conditional marker exposure based on cross-origin isolation

[HeathcliffAC]

feat: Add conditional marker exposure based on cross-origin isolation

Enable selective marker disclosure based on security context:

• COI contexts: all markers (gc, script, paint, layout, style, other)
• Regular contexts: safe markers only (layout, style)

This allows access to layout/style timing (already available via DOM APIs)
in regular contexts while protecting sensitive timing information.

Maintains full backward compatibility with existing API surface.

---

Preview | Diff

[HeathcliffAC]

@arturjanc @ArthurSonzogni I am trying to do a modification on the JS Self Profiler API. Could you help me with any feedback that you could provide?

[ArthurSonzogni]

Thanks! Looping in +@camillelamy as I’m currently on paternity leave and won’t be able to reply confidently immediately from what I recall. This feature dates back to ~4 years ago: here is the previous discussion.

[HeathcliffAC]

@camillelamy It would be great if we could continue the review on this PR. Thanks

[monica-ch]

Hi @HeathcliffAC — I''m picking up this work at Microsoft. Opened #89 which retains your commit and layers a small follow-up on top with three bug fixes (mangled JSON block in markers.md, wrong MSEdgeExplainers URL, collapsed </p><section> in index.html) and a normative filter a marker algorithm that replaces the prose-only conditional-exposure description so the {style, layout} allowlist becomes testable/interoperable. Happy to fold those back into this PR instead if you prefer — let me know how you''d like to proceed. Thanks for the initial draft!