test: stabilize getPendingPosts seed timestamps

[patosullivan]

Summary

OTT, Fixes a flaky getPendingPosts test helper by giving seeded posts monotonic timestamps, avoiding accidental cache-key collisions on fast CI runners.

Noticed after I pushed to #5813 and saw a test failure in CI.

[claude]

Claude Code Review

This repository is configured for manual code reviews. Comment @claude review to trigger a review and subscribe this PR to future pushes, or @claude review once for a one-time review.

_{Tip: disable this comment in your organization's Code Review settings.}

[patosullivan]

@claude review once

[Copilot]

Pull request overview

This PR stabilizes the getPendingPosts test helper by ensuring seeded posts get distinct timestamps, preventing collisions on the posts(cache_id) unique index when multiple inserts occur within the same millisecond on fast CI runners.

Changes:

• Introduces a monotonic seedCounter to make seeded sentAt/receivedAt/syncedAt timestamps unique per insert.
• Updates seeded Post.id generation to align with the computed seed timestamp.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[claude]

LGTM — test-only fix mirroring the existing seedCounter pattern in getDeliveryPendingPosts.

Extended reasoning...

Overview

Single-file test change in packages/shared/src/db/queries.test.ts. Adds a monotonic seedCounter to getPendingPosts's seedPost helper so each seeded row gets a unique sentAt/receivedAt/syncedAt, avoiding cache_id unique-index collisions when inserts land inside the same millisecond on fast CI runners.

Security risks

None. Test-only change; no production code paths, auth, or data handling are touched.

Level of scrutiny

Minimal. This is a flaky-test stabilization that mirrors the exact same seedCounter pattern already in use in the sibling getDeliveryPendingPosts describe block in the same file. The change is small, mechanical, and self-contained.

Other factors

The id template still includes Math.random() so id uniqueness is preserved, and overrides.sentAt ?? t keeps callers' explicit sentAt values (used by tests that pin specific timestamps) intact. No behavior change for the assertions themselves.