Fixes#2054 and #2079.
Packet-row tests could outlive the fixture's default 15-minute window.
An audit of 115 wired suites identified four needing explicit fixture
windows: 180 minutes on mobile, 1440 on desktop. Mobile rejects larger
saved windows, so a desktop-only preference was insufficient.
Packet-icon checks now require a real path-bearing packet and both
detail controls.
Default-window coverage exposed #2079: responsive column hiding also hid
the spanning empty-message cell. Skip structural single-cell spanning
rows when hiding individual columns, preserving status messages and
virtual-scroll spacer height. Ordinary and partial-span data cells
retain their behavior; application defaults stay unchanged.
- Red evidence: `bae5e78` fails the aged-fixture row assertion;
`9b75cf6` fails spanning-row visibility. Guards/setup make them pass.
- Validation: 183 standalone suites; 82 targeted browser cases on a
fixture aged over 30 minutes (one pre-existing node-gesture skip).
Broader browser run: 131 passed, 3 fixture-dependent skips.
- Browser verified: fixture-backed Go server and Chromium;
`coverage/2079-mobile-empty.png`. Real mobile scrolling retains spacer
height and has no horizontal overflow.
- E2E assertions: `test-observer-iata-1188-e2e.js` default-window cases
and `test-table-fluid-e2e.js:58` shared-helper regression, both under
`tests/e2e/`.
- Three independent reviews completed; packet-icon coverage feedback
addressed in `760057d`.
- Populated mobile axe coverage remains part of #1995.
- One constant-time guard per already-visited row; no new requests or
collections.
## Preflight overrides
- External `run-all.sh` unavailable; repository checks run directly.
Local runs use UTF-8, the release harness's `GITHUB_REF_NAME`, and a 60s
navigation budget; repository timeouts are unchanged.
Moves 290 root test-*.js into tests/unit (177, listed in test-all.sh) and tests/e2e (113, classified in scripts/non-unit-tests.json), per #1981 and PR-D of #1385. Root goes from 348 entries to 48. test-all.sh and test-fixtures/ stay put. The inventory guard now fails if a test reappears in the root or sits in the wrong folder.
Verified independently of the diff: the invoked sets are unchanged (test-all.sh 177 before and after, deploy.yml 96 before and after, both identical as sets), and a full local run of test-all.sh on master and on the branch produced 4702 output lines each whose only differences are absolute paths, stack-trace line numbers shifted by the REPO_ROOT line, the inventory wording and two perf ratios. The guard was mutation-checked: a test back in the root, a unit suite in tests/e2e, and a suite dropped from test-all.sh each make it exit 1. CI run 35246304316 ran 97 suites from tests/e2e and is green.
Follow-up 9335c51d finished the instruction files: no bare root test command is left in AGENTS.md, the squad charters, .github or docs, and every tests/ path they name resolves.
Merged by the interim maintainer without a second human reviewer: CI and the local runs above are the independent checks.
Known and deliberately out of scope: 18 of the 113 files in tests/e2e are invoked by no runner at all, and one of them cannot run anywhere because it requires jsdom, which is not a declared dependency. Tracked separately.
n30nex