Ledger Demo Verdict Outcome Resolution

Correspondence-by-observation for the 2026-08-07 verdict_outcome fix (commit 5b9a8ad), per Merlini/Pavlo's proposal (trustless-ai group, topic 16): rather than asking a peer to trust that a gist recompute matches what actually runs in this private repo, call the REAL production resolution function (_resolve_verdict_outcome_citations, the exact code _verdict_outcome_resolution delegates to for real /ledger citations) against a synthetic, clearly-labeled citing set that reproduces the bug report's own scenarios. MUST be registered before GET /ledger/{entry} (same reason as GET /ledger/submit above) -- otherwise the param route would swallow "demo" as an entry id.

No ledger writes happen here -- decision_ref is a synthetic id, citing_entries below are inline literals, nothing is read from or appended to the real /ledger index. This exists purely so the fix's behavior is checkable against live deployed code without either handing over repo access or asking anyone to trust a claim.

This GET route serves ONE fixed example. For your own citation set, POST to this same path with a JSON body ({"citing_entries": [...]}) -- Merlini's honest limit on the fixed version (msg 2431): 'Correspondence is now proven for one input... If the demo took a citation set as a parameter, a reviewer could diff any case they invented, including this one.'