[Bug]: Pull request avatars stay broken when a GHES image cannot load

[tastelessjolt]

Before submitting

• I searched existing issues and did not find a duplicate.
• I included enough detail to reproduce or investigate the problem.

Area

apps/web

Steps to reproduce

1. Connect T3 Code to a private GitHub Enterprise Server instance on a different site from the T3 web client.
2. Open a pull request with activity from users who have avatar URLs.
3. View the pull request activity timeline.
4. The browser rejects the GHES session cookies for the cross-site avatar requests because they use SameSite=Lax or SameSite=Strict.

Expected behavior

If an avatar image cannot load, T3 Code should show its existing actor-initial fallback so the activity timeline remains readable.

Actual behavior

PullRequestActorAvatar only falls back when avatarUrl is null. A non-null GHES URL that fails in the browser remains a broken image in every activity row.

Before	After
Broken images remain in the timeline.	Failed images fall back to actor initials.

Impact

Cosmetic issue

Version or commit

main @ 8ef478e

Environment

T3 Code web or desktop client connected to a private GHES host on a different site. Reproduced in Firefox, where the console reports the GHES authentication cookies being rejected in a cross-site context.

Workaround

No client-side workaround. Opening the pull request directly on GHES keeps the request first-party, but the avatars remain broken inside T3 Code.