Row 38 S3 round 2 gate, the rerun of test-release and test-task. Darkwing.
This adds to r2-gate-summary.txt, which stays as written.

Tree: the same detached worktree at c9ef7ee4 with build-r2.patch applied.
sha256sum -c build-r2-manifest.sha256 (e10e30e3...) gave 28/28 OK before and after the runs. Node v26.8.1.

Limit check, 2026-10-09T12:32:14Z: one live hello task (zai, glm-5.3-flash) in a
throwaway data root succeeded with MOSAIC_HELLO_OK, so the 5-hour limit had reset.

Suites, run one after the other, each teed to its own file:
  start 12:32:22Z, load 1.2
  scripts/test-release.sh   14 passed, 0 failed   done 12:32:33Z
  scripts/test-task.sh      98 passed, 0 failed   done 12:33:06Z

The 11 cases that failed on 2026-10-09 at 00:18Z all pass now:
  test-release: healthy activation succeeds; pointer written with valid fields;
    repeat activation succeeds (log grows).
  test-task: live hello task succeeds with exact marker; result.json contents are
    correct; wrong expectExact fails with exit 1 (expect-mismatch); fork base: teach
    succeeds; fork child recalls ancestor context; forked child recalled ancestor
    code word; user recall run succeeds; recalled user name.

Docker: no case failed on the address pool. The suites used the existing
gate2_default network, and the host's network count stayed at 34. I removed no networks.

These match round 1 (14/14 and 98/98). The 11 failures at 00:18Z came from the zai
usage limit, not from the patch.
