[chriscoveries]

Since 1e5651b2, every pull-request run fails rust-linux on five new fixtures that launch a real Chrome:

• browser_surface::pairing::tests::current_chatgpt_pairing_receipt_keeps_exact_user_and_assistant_boundaries
• browser_surface::prompt_submission::readiness_tests::current_chatgpt_composer_is_not_an_auth_or_search_input
• browser_surface::prompt_submission::readiness_tests::current_chatgpt_layout_submits_once_and_protects_drafts_and_generation
• browser_surface::prompt_submission::readiness_tests::current_chatgpt_user_bubble_is_required_for_delivery_receipt
• browser_surface::prompt_submission::readiness_tests::failed_page_inspection_invalidates_earlier_readiness

Each one stops at the CI guard: "this test launches a real Chrome, which PR CI does not run; add it to the real-Chrome filter in .config/nextest.toml so the nightly job covers it". For example, this PR run for PR #150 at head c3a95c401905aaa5d2974e2afbee34931bd41397 (based on main at 3d3d10cb) fails exactly these five; 2,725 tests pass and 53 are skipped.

This change adds the five names to both filters in .config/nextest.toml, as the file's header asks: the [profile.ci] exclusion and the [profile.chrome] inclusion, each in the lists' existing order. Nothing else changes.

Verification, with GitHub Actions on the fork at the exact head d049fcdf9daad0d519cd0706012153be081a7c1f:

• PR CI, run 39818218409: all six jobs pass, including rust-linux.
• Nightly by workflow_dispatch, run 39818219729. Its browser-surface job ran the five fixtures under the chrome profile against real Chrome; all five passed.