Fix Mike sync Baseline handoff timeout

[ranade-oss]

Root cause

Run 31624069748 passed scan/classification and its Baseline run 31624214386 passed. The failure was the two-hour wait in the low-risk synchronizer: the proposal PR was not merged before the wait expired.

The deeper handoff issue was that the Baseline run was started with workflow_dispatch, but no Handle Baseline result workflow_run invocation was observed for that run. The daily reconciler eventually merged the PR, but only after the synchronizer had already failed.

Fix

• Add a bounded, reusable Baseline settlement helper used by both Mike synchronizers.
• Capture the exact Baseline run by head SHA, wait for completion, and explicitly dispatch the existing Baseline-result gate with that run ID.
• Extend the existing gate with a workflow_dispatch input; it reuses the same exact-head, trusted-PR, review, and mergeability checks and also preserves bounded repair handling for failed Baseline runs.
• Treat delayed merge/reconciliation as a pending successful queue outcome instead of a false workflow failure.
• Suppress low-risk duplicate proposals while an earlier automated proposal is still unmerged.
• Add resolver and workflow-boundary regression coverage.

No new workflow is added; the existing workflow topology is preserved.