Desktop Smoke E2E: root-thread link test flakes at high frequency

## Summary

`desktop/tests/e2e/messaging.spec.ts:2049` (`sends a thread message to its parent channel with a root-thread link`) intermittently times out waiting for the shared parent-channel row after closing the auxiliary panel.

## Evidence

- PR #6590 run `32901731027`, job `97977094376`, head `90f046d53`: failed all 3 attempts at the 5s predicate (`messaging.spec.ts:2228-2241`). Final screenshot showed the expected shared row present, consistent with a timing race.
- The unchanged-head failed-job rerun, job `97983581661`, passed.
- The same test previously failed 3/3 at `e7220100e`, then passed first-try at `90ffff85c` and `fef7aee18` on trees containing the relevant channel UI.

This is therefore a high-frequency per-run flake rather than a deterministic PR regression, but 6 failed attempts across two runs warrants root-cause work instead of relying on retries.

## Suggested investigation

Trace the transition from `auxiliary-panel-close` to parent-channel rendering and replace the polling race with an observable readiness condition (or fix the underlying state propagation delay). Preserve the test's assertion that the sent-from-thread message appears in the parent channel with the correct root-thread link.

Start in desktop/tests/e2e/messaging.spec.ts around line 2049 and inspect the wait at lines 2228-2241, then trace the transition after auxiliary-panel-close. Replace the polling race with an observable readiness condition or address the state propagation delay, while preserving the assertion that the sent thread message appears in the parent channel with the correct root-thread link.

Written by the indexing model from the issue text.