perf(threads): virtualize the thread panel's replies - #2911

## Summary

These are fixes for three browser-journey flakes that are failing open
PRs which don't touch the code under test. The change is test-only. No
browser cases are added or removed, and no assertion is relaxed.

The red `Browser journeys (chromium, 3/9)` / `(webkit, 4/14)` shards on
most open PRs come from a different problem:
`sidebar-geometry.spec.mjs:224` is broken on main. #2945 and #2946
already fix it, so this PR leaves it alone.

### 1. `channel-completion.spec.mjs`: cold dev-server load timed out

**Evidence:** `Flake check` on #2935 failed
`channel-completion.spec.mjs:7` on both engines with `page.goto: Timeout
15000ms exceeded` on
`/tests/fixtures/mentions.html?channels&stream&test-controls`
([chromium](https://github.com/block/buzz-app/actions/runs/42317084302/job/128232679087),
[webkit](https://github.com/block/buzz-app/actions/runs/42317084302/job/128232678918)).
#2935 doesn't touch that page.

**Cause:** this is the same cold load #2776 fixed for
`mentions.spec.mjs`. Each test here started its own Vite dev server for
the same ~1,600-module fixture.

**Fix:**
- `beforeAll` builds `mentions.html` once and serves it with
`build`/`preview`, using the same config as `mentions.spec.mjs`.
- The build keeps the dev server's CSS. Tailwind's optimizer and Vite's
CSS minifier each round the 20/14 line-height token to `1.42857`, so
line boxes become `19.99998px`. In WebKit that put the popup `4.03125px`
above the composer, and the test asserts exactly `4`. I confirmed this
locally: it fails 5/5 with either setting left on and passes with both
off. Turning both off keeps the geometry the test checked before.

### 2. `profiles.spec.mjs:940`: reopening Members right after closing
the profile

**Evidence:** `Flake check (webkit, 8/9)` on #2909 failed in repeat 8 of
10
([job](https://github.com/block/buzz-app/actions/runs/42313241314/job/128219877861)).
The trace shows this sequence:
1. `Escape` closes the profile panel.
2. Focus returns to "Manage members of One" within 40ms.
3. `Enter` is pressed 10ms later.
4. The DOM after that `Enter` shows the Profile panel again and no
Members dialog, so the next `row.click` times out.

**Fix:** after each `Escape`, wait for the address to leave the profile
(no `"detail"` in the `#buzz=` hash) before reopening Members. This is a
lifecycle barrier, not a delay. One step in this test already waited for
the panel to unmount before reopening; the other steps didn't wait at
all.

**Product follow-up (not fixed here):** the order of a dismissal's
history return and a quick reopen matters.
- If I delay `history.go` by 300ms, the unpatched test takes a different
path: the dismissal's `go(-1)` never runs, and the URL stays on the
profile visit under the reopened dialog.
- A `browser-history` unit sketch (deferred traversal: `push(channel)`,
`push(panel)`, `dismiss([panel])`, then `push(settings)` before the
traversal commits) lands on `channel`. The new `settings` visit is
undone.

Someone who owns navigation should decide whether a push during a
pending dismissal return should be deferred or replayed. That is a
navigation-history change, not a test fix.

### 3. `timeline.mjs` `edge()`: one wheel gesture stopped short

**Evidence:** `Browser journeys (webkit, 6/14)` on #2937 failed
`history-prefix.spec.mjs:37` with `wheel reaches timeline edge`,
`Received: 199`
([job](https://github.com/block/buzz-app/actions/runs/42320781857/job/128244819852)).
The trace shows the one `-3251px` wheel ended with `scrollend` and the
geometry then held still, so the poll could never pass.

**Fix:** `edge()` repeats the gesture, at most four times, until the
edge is reached. Every extra gesture must reduce the distance, so an
immobile timeline still fails. `upper()` already follows this rule. The
case of already being at the edge is unchanged: it still sends one small
wheel so production can start paging.

## Not in this PR, from the same survey of the 40 most recently updated
open PRs

- `sidebar-geometry.spec.mjs:224` (Drafts & sent row) is broken on main
since #2733 + #2728 crossed. #2945 is green and #2946 is in CI; either
one fixes it.
- `solo-chats.spec.mjs:322` (Share, webkit, on #2927) is being handled
in draft #2933.
- `layout.spec.mjs:803` on #2916 is already fixed on main by #2750.
#2916's merge ref predates it.
- #2911 (thread virtualization) and #2918 (per-channel notifications)
fail consistently in tests their own changes reach, so those look like
regressions for their authors, not flakes.

## Validation

All runs were local on macOS with
`tests/browser/playwright.ci.config.mjs` and `--no-deps`. These are not
hosted CI measurements.

- `channel-completion.spec.mjs`, `--repeat-each=5`, both engines: 20/20
passed with the change. Before the CSS settings were added, the built
fixture failed WebKit 5/5 (`4.03125`).
- `profiles.spec.mjs:940`, `--repeat-each=24 --workers=6`, WebKit:
passed. It also passed with `history.go/back/forward` delayed by 300ms,
in both engines.
- `history-prefix`, `history-loading`, `scroll`, `timeline-setup` and
`new-message-boundary` (all callers of `edge()`/`end()`), both engines:
70/70 passed.
- Full `history-prefix`, `channel-completion` and `profiles` files, both
engines: 43/44 passed. The one failure was `profiles.spec.mjs:6` in
WebKit, a test this PR doesn't touch. It then passed 4/4 both with and
without this change.

**Deferred and remaining gaps:**
- I couldn't reproduce the cold-load timeout or the WebKit stopping
short locally, so I have no local fail-before for those two. The
evidence is the CI traces linked above.
- The other `edge()` callers (24 spec files) are left to CI.
- A green run doesn't prove these flakes are gone.

Signed-off-by: Bradley Axen <[email protected]>
Co-authored-by: mbot <3754f8729004d95654c46dbab3129e4ab9ef05cc2534e2a3fbfc155983bd637b@buzz.block.builderlab.xyz>

…tivity and message-navigation (#3444)

## Why

Two journeys fail intermittently in CI because each waits on a signal
that does not match what the test then asserts.

| Test | Sightings |
|---|---|
| `local-agent-activity.spec.mjs`, "opens an active message's empty
thread without losing its activity preview" | Browser journeys: [10 Nov,
main, WebKit
18/18](https://github.com/block/buzz-app/actions/runs/42849358399/job/129946045226),
[10 Nov, Chromium
9/9](https://github.com/block/buzz-app/actions/runs/42856816690/job/129972535241),
[11 Nov, WebKit
18/18](https://github.com/block/buzz-app/actions/runs/42933561357/job/130238403731),
[11 Nov, WebKit
18/18](https://github.com/block/buzz-app/actions/runs/43077988192/job/130723357985)
|
| `local-agent-activity.spec.mjs`, "finishing activity retires keyboard
controls without stealing focus: stop" | Browser journeys: [10 Nov,
WebKit
18/18](https://github.com/block/buzz-app/actions/runs/42840658251/job/129915569261)
|
| `message-navigation.spec.mjs`, "exact reveal uses ordinary dwell
rather than marking read at open" | Browser journeys: [10 Nov, main,
Chromium
5/9](https://github.com/block/buzz-app/actions/runs/42922432651/job/130200813397)
|

## Root cause and fix

One commit per spec.

**Thread rows still positioning.** While a thread positions its first
page, `ThreadPanel` renders its rows with `inert` and sets `aria-busy`
on the "Thread messages" region. Playwright's `toBeVisible()` ignores
`inert`, and `locator.press()` and `focus()` make no actionability
check. The two steps added in #3140 could therefore run before
positioning finished.

- In the empty-thread test, `activity.press("Enter")` went to the
still-focused reply composer, which is prefilled with a mention. WebKit
sent the draft and raised the "Mention people outside this channel?"
dialog. Chromium kept focus in the composer. The Activity tab never
appeared. The composer's `toBeFocused()` passes early because the
composer sits outside the inert rows.
- In the "stop" variant, `focus()` on Stop was a no-op, so the handoff
to "More message actions" never happened.

DOM snapshots from the four traces examined show `aria-busy="true"` and
`inert` on the rows at the moment of the press or focus. Every sighting
post-dates #3140. This is not a product bug: people cannot reach inert
rows either.

The tests now wait for the region's `aria-busy="false"`, the readiness
signal `channel-lifecycle`, `nested-replies` and
`navigation-thread-history` already use. The wait sits after the
composer `toBeFocused()` and before the Enter press, and in the
finishing loop after the bubble is visible and before any `focus()`, so
the bubble variant is covered as well.

**Read publication under a default timeout.** The exact-reveal dwell
test ended by polling the fixture's read publications with the default
10 s timeout. A publication needs the 300 ms reading dwell, then
read-state's 5 s publish debounce, then the flush's read, sign and
publish. That leaves about 4.5 s of slack. In the failing trace the
flush's read reached the fixture on time and the relay answered in about
4 ms, but the browser took 5.6 s to receive the response body. The
signing request arrived 0.9 s after the poll deadline. The evidence file
at teardown held the publication, so the behaviour was correct. The test
was waiting on the wrong signal for what it asserts, which is that dwell
marks read and open does not.

The test now asserts the local read with `readJournal` from
`reading.mjs`, as the sibling test in the same file does. Right after
the held query is seen, the journal's frontiers are still empty. At the
end it polls for a non-empty frontier set. The journal write is strictly
earlier than the publication and does not depend on the relay
round-trip. The existing check that publications are unchanged at open
is kept. The publication path stays covered by `unread.spec.mjs`, which
polls with a 12 s timeout.

No timeouts or tolerances changed. No browser cases were added or
removed.

## Evidence

Local runs on Linux, `bin/pnpm test:browser:ci --no-deps
--reporter=list`.

| Scope | Result |
|---|---|
| `local-agent-activity.spec.mjs -g "opens an active message\|without
stealing focus"`, WebKit, `--workers=1 --repeat-each=8` (4 tests × 8) |
32/32 passed |
| Same, Chromium | 32/32 passed |
| `message-navigation.spec.mjs -g "exact reveal uses ordinary dwell"`,
Chromium, `--workers=1 --repeat-each=8` | 8/8 passed |
| Same, WebKit, `--repeat-each=4` | 4/4 passed |
| Whole `message-navigation.spec.mjs`, `--workers=2` | Chromium 22/22,
WebKit 22/22 |

During triage, a test-only route that delayed the relay query by 1.5 to
2.5 s reproduced the inert-row failure deterministically: without the
wait WebKit and Chromium each failed 2/2 with the CI signature, and with
the wait all passed.

## Not reproduced, not run

- Neither failure reproduced on an idle host without that injected
delay. The slow response body in the read-publication failure was pinned
to CI load. The repeat runs guard against regression rather than
reproducing CI timing.
- No human has tested this yet, so `buzz-review-completed` is not set.

## Left alone

- The same default-timeout publication poll remains in "settings blocks
reading and focus in a retained thread" in this file and in
`notifications.spec.mjs`.
- Open PRs #3183, #3001 and #2610 edit other tests in
`local-agent-activity.spec.mjs`. Open PRs #2911, #2563, #2555, #2407 and
#2401 edit other parts of `message-navigation.spec.mjs`. A trial merge
of each against this change is clean.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Signed-off-by: mtoohey <[email protected]>
Signed-off-by: Matt Toohey <[email protected]>
Co-authored-by: Claude Fable 5.1 <[email protected]>