P10.9B: qualify full BRCB receive chronology and non-GI post-idle recurrence

[masarray]

Evidence: operator BCUGE 2026-10-11 18:22 / 18:24 WIB

Two private full ARSAS Diagnostic Reports from merge-test executable 1.6.40+7bc0241f6361474a4e63f748be18097f52b84a2b (GitHub-verified merge SHA with parent P10.9A a27d108b...). Locked engine 00ad2819b99ffe6b3f885e59aa24a6afb2b4d8e1. Open SCL/Discovery have matching static semantic fingerprint 439d4dc0507019533b154ed68f7f4007f39f6be92b5cd3f413ff3b84f619727e, 8 DataSets / 124 members, 67/67 visible points, 6/6 static RCBs actually routed, 0 cyclic MMS polling. Different BRCB slots remain: Open SCL CSWI/EVE GW vs Discovery CSWI/EVE SOGI; no valid paired reliability A/B.

Open SCL: 49 frames / 16 findings / 13 SqNum discontinuities / 3 BufOvfl; CSWI 39/39 bounded chronological receive frames with SqNum 12→1, GIreason=NO at 11:22:13.8183266Z, ~20.042s after activation return and 19.734s after preceding observed frame; only one later clean report. Discovery: 50 frames / 14 findings / 11 discontinuities / 3 BufOvfl; CSWI 40/40 bounded chronology with SqNum 8→1, GIreason=NO at 11:24:01.3554255Z, ~19.905s after activation return and 19.615s after preceding frame; only one later clean report. EVEGGIO only two initial GI+BufOvfl frames and no follow-up in both sessions. This is stronger evidence against a startup-only observational pattern, NOT proof of packet ordering, server/client cause, event loss or physical R10. Raw private diagnostics were not uploaded to repository.

P10.9B isolated offline enhancement

Extends existing P10.9A opt-in --independent-triage mode only, with BRCB bounded decoder receive-order completeness gate:

• Only when declared first+recent chronology covers 1..N frames without omitted middle, and every decoder timestamp is valid/monotonic, calculate maximum observed receive gap.
• Surface max gap preceding a non-GI, non-BufOvfl, backward/repeat SqNum frame with an actual diagnostic finding, including adjacent frame IDs, SqNum values, timestamps, from-RCB-start timing. Described as received-frame observation, never a proven reset/replay or SOE loss.
• Missing, truncated, duplicate/misordered, inconsistent, untimed, offset-naive or nonmonotonic history → UNVERIFIABLE with explicit reason and null gap/event; never substitute zero or extrapolate wire order.
• Default P10.8E strict same-slot compare() contract unchanged; earlier P10.9A independent per-session records maintained. No new CI workflow, no device interaction, no engine/runtime/RCB write/decoder/GI/control/polling changes.
• Original 19 Python tests plus 6 new deterministic positive/negative tests = 25 tests, canonical Windows workflow runs them. No raw operator logs committed.

Run python scripts/compare-p10-brcb-sessions.py OpenSCL-Diagnostic.txt Discovery-Diagnostic.txt --independent-triage --output ReceiveEvidence.json.

Preserve immutable P10.6, P10.8E, P10.9A checkpoints. Branch based exactly on a27d108b765cb725f042c85b584b254969a56519, engine lock 00ad2819... unchanged; #512 stays draft, no main merge or public release. Promote only after all exact head CI succeeds. #520 remains open for real wire capture and same-slot field validation.