G2.6 P1.6: restore field-proven general dynamic RCB runtime

[masarray]

Goal

Restore the original Smart Auto runtime contract for an exact identity/profile whose dynamic reporting mechanism has already been physically proven: static reporting first, then bounded dynamic RCB/DataSet coverage for all still-uncovered exact-resolved selected signals, then MMS only for genuine residuals.

Why

P6.2-B deliberately quarantined automatic dynamic activation after advertised capability alone proved insufficient. Later G2.4/G2.5/P1.5 field work now provides stronger physical evidence: a bound NO-GI spontaneous dchg InformationReport, exact member mapping, healthy association, cleanup, and reconnect/re-arm. P1.5b correctly made that witness usable but incorrectly kept the witness as the permanent runtime member scope.

P1.6 treats the physical Q0/A3 chain as a capability witness, not a universal member whitelist.

Safety retained

• exact stable identity/model fingerprint/profile revision binding;
• physical dchg witness must remain an ordered subset of the persisted successful qualification chain;
• static DataSet coverage keeps precedence;
• dynamic RCBs must be freshly exact-verified empty/free on the current association;
• dynamic members must resolve exactly in the current live MMS directory;
• normal planner MaxDynamicReportPlans / MaxDynamicMembersPerReport bounds remain in force;
• no profile mutation/save;
• no ProductionEligible authorization or synthesis;
• callers retain runtime cleanup/reconnect/circuit-breaker enforcement.

Regression coverage

• witness has only two members but five exact-resolved selected signals become dynamic across multiple bounded URCB groups;
• identity mismatch fails closed;
• missing cleanup evidence fails closed;
• busy dynamic slot is not used and residual points poll;
• member absent from live directory remains polling residual;
• ProductionEligible remains false.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6dec7ff5fb

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Require exact confidence for dynamic members

ValidateDynamicSegments treats every positive-confidence directory point as “exact.” When the effective directory contains supplemental points from LiveIedModelDiscoveryBuilder.BuildEffectiveDirectory—which preserves member confidences such as 50 or 90—a matching requested signal can therefore produce an active dynamic DataSet write despite not having exact resolution. Require genuinely exact live-directory evidence rather than Confidence > 0, or keep these signals on polling.

AGENTS.md reference: AGENTS.md:L140-L147

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Document the new runtime capability and evidence level

This commit exposes a completed P1.6 runtime capability but updates none of the required user scope, maturity, changelog, or validation documentation, leaving consumers and assurance tracking unable to distinguish this new field-capability behavior from the previous subset-scoped runtime. Update the appropriate documentation alongside the implementation.

AGENTS.md reference: AGENTS.md:L115-L124

Useful? React with 👍 / 👎.