Stabilize non-fast-forward fallback tests in push_to_pull_request_branch (CJS shard 4)

[Copilot]

JS Tests (shard 4/4) was red on every push/PR due to two regressions in push_to_pull_request_branch.test.cjs around non-fast-forward handling. The failures came from brittle ordered mocks that no longer matched the expanded git call sequence in pushSignedCommits.

• ### Root cause captured in tests

  • The two fallback-path tests assumed a strict mockResolvedValueOnce order for exec.getExecOutput.
  • Recent signed-push flow changes added intermediate git calls (rev-parse, rev-list --parents, diff-tree), shifting those assumptions and invalidating assertions.

• ### Test harness updated (minimal scope)

  • Replaced ordered mockResolvedValueOnce chains in the two failing cases with command-shape-based mockImplementation.
  • Kept scenario intent unchanged:
    • default behavior creates fallback PR on non-fast-forward
    • fallback_as_pull_request: false returns push_failed and does not create fallback PR

• ### Behavior preserved

  • No production code changes; only test setup logic was made resilient to internal call-sequence evolution.

mockExec.getExecOutput = vi.fn().mockImplementation(async (_cmd, args) => {
  const a = Array.isArray(args) ? args : [];
  if (a[0] === "rev-list" && a[1] === "--parents") {
    return { exitCode: 0, stdout: "2222... 1111...\n", stderr: "" };
  }
  if (a[0] === "ls-remote" && a[2] === "refs/heads/feature-branch") {
    return { exitCode: 0, stdout: "1111...\trefs/heads/feature-branch\n", stderr: "" };
  }
  return originalGetExecOutput(_cmd, args);
});

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.