fix(demo): gate invite-path RM triage on Finder having case replica (Bug #2120)

[sei-ahouseholder]

• Closes Bug: fcvcv and fvcv-handoff demos fail — ReconstructChainTail cannot bootstrap unanchored ledger chain (CLP-08-005) #2120

Summary

run_invite_path_rm_triage in fcvcv_demo.py and fvcv_handoff_demo.py triggered RM state updates that broadcast Announce(CaseLedgerEntry) to the Finder before the Finder's VulnerabilityCase (genesis hash) had been seeded, causing ReconstructChainTailNode to raise CLP-08-005 (unanchored chain bootstrap) in the fcvcv and fvcv-handoff demo scenarios.

Changes

• vultron/demo/scenario/fcvcv_demo.py: Add wait_for_case_on_container(finder_client, case.id_) before V1 triage in _phase_report_submission; add finder_client parameter to _phase_c2_suggests_v2 and call site; add finder wait before V2 triage
• vultron/demo/scenario/fvcv_handoff_demo.py: Add finder_client parameter to _phase_coordinator_invites_vendor2 and call site; add finder wait before Vendor2 triage
• test/demo/test_fcvcv_demo.py: New file — 3 regression tests verifying finder-wait-before-triage ordering invariant (spec CLP-08-005)
• test/demo/test_fvcv_handoff_demo.py: 2 new regression tests for _phase_coordinator_invites_vendor2 ordering invariant

Verification

• 5 new regression tests pass (3 in test_fcvcv_demo.py, 2 in test_fvcv_handoff_demo.py)
• Full unit suite: 6467 passed
• Black, flake8, mypy, pyright clean
• 4 pre-existing integration failures confirmed on base branch (test_integration_script_scenarios CI config mismatch, test_pcr_bootstrap test isolation — both pre-date this PR)

🤖 Generated with Claude Code

[sei-ahouseholder]

PR Triage: #2127 — fix(demo): gate invite-path RM triage on Finder having case replica (Bug #2120)

Linked issues: #2120 (Bug: fcvcv and fvcv-handoff demos fail — ReconstructChainTail CLP-08-005)
Changed files: 5 — demo/scenario, test/demo, plan/history
CI status: ❌ failing (pre-existing only — see notes)
Needs integration tests: yes (5 regression tests added)

---

CI Summary

All failing CI jobs are pre-existing on main (confirmed against run 31233723925):

Job	Failure	Pre-existing?
Tests (pytest)	test_pcr_bootstrap (#2086), test_integration_script_scenarios (#2122)	✅ yes
fcvcv Demo Integration	engage-case 422 / RM.RECEIVED timeout	✅ yes (different error from #2120)
fvcv-handoff Demo Integration	Ownership-transfer offer delivery timeout	✅ yes (different error from #2120)
fcvcv / fvcv-handoff Invariant Harness	Downstream of demo failure	✅ yes

All linters (black, flake8, mypy, pyright) pass.

---

Findings

#	Phase	Severity	Description	Outcome
phase8-unguarded-triage-sibling-scenarios-0	code-review	❌ FAIL	5 sibling scenario files call run_invite_path_rm_triage without wait_for_case_on_container(finder_client) guard — same CLP-08-005 defect	new-issue-ask

Total: 1 FAIL · 0 IMPROVE · 0 NEW-ISSUE-NO-ASK

---

Finding Detail

phase8-unguarded-triage-sibling-scenarios-0 — CLP-08-005 defect in sibling scenarios

The fix is correct and complete for the two scenarios named in #2120. However, the same root cause (Finder's genesis hash not seeded before run_invoke_path_rm_triage fires) exists in 5 other scenario files:

File	Line	Phase function	finder_client param?
fccv_handoff_demo.py	538	_phase_c2_invites_vendor	❌ not a param
fvcv_extension_demo.py	425	_phase_coordinator_suggests_vendor2	❌ not a param
fccv_extension_demo.py	427	_phase_c2_suggests_vendor	❌ not a param
fvv_demo.py	270	_phase_report_submission	✅ in scope
fcv_demo.py	330	_phase_invite_vendor	❌ not a param

This is non-trivial (requires threading finder_client through 4 call chains) and distinct from #2120's stated scope, so it warrants a new tracking issue.

User decision needed: Should these be fixed in this PR (expands scope to all 6 scenarios), or tracked in a new issue to stack on top?

---

Triage artifact: .claude/pr-2127-triage.json
Next step: run /pr-ship 2127 after user responds to the scope question above.

[sei-ahouseholder]

Branch freshened onto current main (4045a53) via cherry-pick. Resolves the pre-existing test failures from the old CI run:

• test_integration_script_scenarios — was broken on the old base, fixed by fix: point scenario ratchet at demo-scenarios.json #2123 which has since merged
• test_pcr_bootstrap — was broken on the old base, fixed by fix(test): stop demo config cache leaking across tests #2126 which has since merged

Full unit suite (6453 passed) and all demo tests (1055 passed) confirm clean on the freshened branch.

[sei-ahouseholder]

Closing as superseded

Every change this PR would merge is already present on main. Code-level evidence:

Production code — identical diffs already on main

Commit 9fad5aa1 (fix(demo): gate invite-path RM triage on Finder having case replica) is a byte-for-byte match to this PR's changes in fcvcv_demo.py and fvcv_handoff_demo.py:

• wait_for_case_on_container(finder_client, case.id_) inserted before V1 triage in _phase_report_submission
• finder_client parameter added to _phase_c2_suggests_v2; same wait inserted before V2 triage
• finder_client parameter added to _phase_coordinator_invites_vendor2; same wait inserted before Vendor2 triage
• All three call sites in run_fcvcv_demo / run_fvcv_handoff_demo updated

A second commit, bfde8167 (fix(demo): add genesis-level Finder case-replica wait in _phase_report_submission, closes #2120), added an earlier genesis-hash-level wait in both scenario files — coverage this PR does not include.

Tests — main has a superset

File	This PR	main
test/demo/test_fcvcv_demo.py	293 lines (new)	405 lines
test/demo/test_fvcv_handoff_demo.py	+133 lines	already contains TestFinderCaseReplicaWaitBeforeVendor2Triage + additional genesis-level tests

All three test classes this PR introduces (TestFinderCaseReplicaWaitBeforeV1Triage, TestFinderCaseReplicaWaitBeforeV2Triage, TestFinderCaseReplicaWaitBeforeVendor2Triage) are present on main, with additional coverage beyond what this PR provides.

Closing in favour of the work already merged.

[sei-ahouseholder]

Superseded — see comment above for full code-level evidence.