Deflake test_memo: gate on hydration and capture page diagnostics

[masenf]

test_memo interacted with the page immediately after goto, racing the
websocket connect/hydrate startup; the typed input event could be lost,
leaving #memo-last-value empty through every rerun of an affected CI
job. Gate all memo tests on the standard #token hydration marker used
by the rest of the playwright suite before interacting.

Also add a tests_playwright conftest that records console messages,
page errors, and websocket frames per test and attaches them to failed
test reports, so any future flake of this kind is diagnosable directly
from CI job logs instead of reproducing blind.

https://claude.ai/code/session_01Xvaqxi4LJeWu2fo4Fk7qMD

[greptile-apps]

Greptile Summary

This PR fixes a race condition in test_memo where tests interacted with the page before the WebSocket/hydrate cycle completed, causing typed input events to be silently dropped in CI. It also introduces a new conftest.py for the Playwright suite that records console messages, page errors, and WebSocket frames per test and attaches them to failed reports.

• test_memo.py: adds a #token input bound to router.session.client_token and a load_page() helper that waits for it to be non-empty before any test interaction, using the same hydration-gate pattern as the rest of the Playwright suite.
• conftest.py (new): autouse _page_diagnostics fixture hooks into Playwright's page events and a pytest_runtest_makereport hookimpl attaches the captured log to failed/rerun call reports.

Confidence Score: 4/5

Safe to merge; the hydration fix is correct and the diagnostic conftest is a net improvement, with one minor gap in the WebSocket frame logging.

The WebSocket frame callbacks pass the raw WebSocketFrame object into an f-string rather than its .payload attribute, so frame contents will appear as opaque object addresses in CI logs. This doesn't affect test correctness — the hydration fix is solid — but the diagnostic output that motivated the conftest will be less useful than intended whenever a WebSocket-related flake occurs.

tests/integration/tests_playwright/conftest.py — the ws frame logging callbacks on lines 44-45.

Important Files Changed

Filename	Overview
tests/integration/tests_playwright/conftest.py	New conftest adding autouse diagnostic capture for console/pageerror/websocket events; WebSocket frame callbacks log the raw frame object instead of frame.payload, producing opaque addresses in CI logs.
tests/integration/tests_playwright/test_memo.py	Adds hydration guard via #token input and load_page() helper; all five tests updated consistently; no issues found.

_{Reviews (1): Last reviewed commit: "Deflake test_memo: gate on hydration and..." | Re-trigger Greptile}

[greptile-apps]

The frame argument passed to framesent/framereceived callbacks is a Playwright WebSocketFrame object. It does not override __repr__ or __str__, so interpolating it directly with an f-string logs an opaque object address (e.g. <WebSocketFrame object at 0x7f…>) rather than the actual payload. Use frame.payload to capture the frame contents that make these diagnostics useful.

Suggested change

	ws.on("framesent", lambda frame: stamp(f"ws sent: {frame}"))
	ws.on("framereceived", lambda frame: stamp(f"ws recv: {frame}"))
	ws.on("framesent", lambda frame: stamp(f"ws sent: {frame.payload}"))
	ws.on("framereceived", lambda frame: stamp(f"ws recv: {frame.payload}}"))

[codspeed-hq]

Merging this PR will not alter performance

✅ 26 untouched benchmarks
⏩ 8 skipped benchmarks¹

---

_{Comparing claude/intelligent-bell-2sg4ti (8476246) with main (fdb00db)}

Footnotes

1. 8 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[masenf]

this actually doesn't seem like it would fix the problem