fix: exact maintainer status check with negative-cache lockdown (#871)

[ionfwsrijan]

Problem

isUserMaintainer fetched only 20 installation users (.limit(20)) then filtered client-side, so a maintainer beyond the first 20 could be misclassified as a non-maintainer. Worse, a transient failure when the service client was missing cached a permanent "denied" answer, locking legitimate maintainers out.

Fix

• src/lib/maintainer/detect.ts: the DB query now uses an exact EXISTS filter (.eq('github_installations.uninstalled_at', null).limit(1)); the client-side .some() truncation is gone.
• A missing service client no longer writes a negative cache entry.
• The negative cache is explicitly busted (cacheDel) during maintainer discovery and when an installation event (re)creates an install.

Files changed

• src/lib/maintainer/detect.ts
• src/inngest/functions/maintainer-discover.ts
• src/inngest/functions/process-installation-event.ts
• Test: detect.test.ts

Testing

• Full MergeShip suite passes (tests, tsc, eslint).
• New tests cover a maintainer at position >20 and the no-cache-on-missing-service case.

Closes #871

[vercel]

@ionfwsrijan is attempting to deploy a commit to the codersogs-3057's projects Team on Vercel.

A member of the Team first needs to authorize it.

[github-actions]

Hey @ionfwsrijan

You have 4 open PRs right now. The limit is 3 at a time.

Please get your existing PRs merged or closed before opening new ones:

• fix: reconcile merged-PR XP across all installs and pages (#872) #877 - fix: reconcile merged-PR XP across all installs and pages ([BUG] reconcileMergedPrXp never scans beyond the first 50 installs, first 20 repos, or the first 100 closed PRs — lost merge XP is permanently unrecoverable on busy repos #872)
• fix: maintainer queue pagination with in-SQL filters and exact counts #868 - fix: maintainer queue pagination with in-SQL filters and exact counts
• fix: bound issues-sweep cost with per-repo checkpoints and cooldown  #865 - fix: bound issues-sweep cost with per-repo checkpoints and cooldown

This PR will remain open but won't be reviewed until you're under the limit. See our Contributing Guidelines for details.

[ionfwsrijan]

@jakharmonika364 @Soumya-codr Please review this

[jakharmonika364]

The core .eq(..., null) -> should-be .is(..., null) bug is a one-line fix but is severe: as written, this PR would make isUserMaintainer return false for 100% of users in production, a total maintainer-detection outage. Everything else in the diff (cacheDel cleanups) is correct and can stay as-is once that line is fixed.