test(packets): verify observation hex after rerenders

[n30nex]

Red commit: ca553e4 (expected assertion failure).

Fixes #2104.

The observation hex test now opens the existing grouped fixture directly, selects A → B → A with stable-ID locators, and asserts the selected ID and exact raw bytes after each render. Missing fixture data fails explicitly. Distinct observation frames are seeded after schema migration; this test's fixed sleeps and silent skips are removed.

Rebased onto master 415362c after #2090 merged. Both patches are unchanged. The previous #1122 layout blocker now passes against the actual current server and assets.

Validation at 5db8017:

• All 185 standalone frontend suites passed.
• Core browser: 133 passed, three existing skips; bug(packets): filter UX disaster — help panel overlaps table, toolbar chaotic, path chips spill rows #1122 layout 6/6, filter 11/11, grouped-collapse 4/4.
• Syntax, YAML, fixture SQL, XSS and lint passed (zero errors, 92 existing warnings).
• Fresh red CI reached the intended detached-observation assertion. Green CI passed; three independent automated reviews found no must-fix issues.

Only test code and fixture setup change; no configuration/customizer implications.

Preflight overrides

External OpenClaw runner/profile unavailable; repository checks and actual local Chromium were used.

[n30nex]

CI status at 9135520: the new observation regression passed. The later #1122 row-height test failed on existing table sizing.

I reproduced identical failures with this PR's fixture bytes restored to their old values. Paths remain 22px tall; an oversized empty Expand column squeezes Details and grows whole rows.

The unchanged test passes 6/6 with #2090's three frontend assets at 541bde6, on the same API/data (maximum row 211→41px). That verifies this blocker only.

Keeping this draft pending the layout fix, then final CI and independent review. The assertion remains intact.

[efiten]

@n30nex #2090 is merged (a4a447ac), so the packets layout fix this PR was waiting on is now in master. Could you rebase onto current master? The #1122 row-height test should pass after that, and then this can come out of draft for final CI and review.

[n30nex]

Thanks — I’ve confirmed #2090 is in current master. I’m rebasing this branch now, then rerunning the observation regression, #1122 layout check and the relevant browser suites. I’ll preserve the test-first history and keep it draft until fresh CI and independent review are complete.

[n30nex]

Review feedback addressed (commit 5db8017):

1. Rebased onto 415362c after fix(packets): pack short columns, one expand arrow, Full Names toggle #2090; both patches remain unchanged.
2. The bug(packets): filter UX disaster — help panel overlaps table, toolbar chaotic, path chips spill rows #1122 layout check now passes. All 185 standalone suites, the core browser suite (133 passed, three existing skips), and related filter/group checks passed locally.
3. Fresh CI passed Go, full Playwright and Docker. The rebased red checkpoint still fails on the intended detached-row assertion.
4. Three independent automated reviews found no must-fix issues.

Ready for maintainer review.