test(review): regression test locking advisory==live verdict/disposition parity

[JSONbored]

Formal regression test locking the advisory==live parity verified in #1569's code audit (processors.ts:768-769, 946-947, 1143, 1206, 1233): advisory mode must compute the byte-identical verdict/disposition as live mode, differing only in whether the merge/close action actually executes.

Deliverables

• Add a test that runs the review pipeline in both advisory and live autonomy modes over the same fixture PR set and asserts the computed verdict/disposition/confidence are identical
• Assert the only difference is that live mode calls the merge/close action and advisory mode does not (mock/spy the action call)
• Cover at least one PASS and one FAIL/blocked disposition case
• 97%+ branch coverage on the compared code paths

References

• src/queue/processors.ts:768-769
• src/queue/processors.ts:946-947
• src/queue/processors.ts:1143
• src/queue/processors.ts:1206
• src/queue/processors.ts:1233

Split from #1569 (closed as functionally complete; this is the residual test-only gap).

---

size: S · gittensor:bug

[linear-code]

JSO-2291