fix(discovery): preserve canonical static DataSet authority after live model build

[masarray]

Physical regression addressed

The invalidated discovery-static candidate exposed two separate regressions.

1. Discovery provenance / entry-point regression

R10 physical acceptance was performed on the Smart Discovery field artifact. That workflow ran scripts/enable-smart-discovery-capture.ps1 before compilation, which injected the proven DiscoverSignalsAsync -> DiscoverSignalsSmartForCaptureAsync route plus association-generation reset/control-authority wiring.

The repository commit itself did not contain that route. A normal build from the same commit therefore executed the older DiscoverSignalsAsync path: supplemental GetNameList/type/probe/unit work returned, explaining why the later normal build had slow discovery and lacked the SMART-CAPTURE ... TimingMs ... diagnostic marker.

This PR promotes that physically proven build transform into normal source and records the provenance correction in the R10 evidence.

2. Live-model Static DataSet projection regression

After model creation, the direct live static path also diverged from the proven opened-SCL path:

• it re-read the DataSet directory and replaced canonical member identity/order with that raw projection;
• it used a different generic/client-compatible activation entry point;
• the routing-only PR [INVALIDATED] fix(discovery): reuse static DataSet workflow after model build #332 exposed this latent path and physical testing showed 46 runtime rows from 58 members, repeated Boolean/quality BIT STRING misprojection, and only 11 report-backed points.

Repair

• normal source now enters the physically proven Smart Discovery single-flight route;
• reconnect/dispose invalidates Smart Discovery association generation in source, not only in a build-time patch;
• control inspection reuses the authoritative smart domain inventory;
• discovery completion enters the shared DataSet Signals / Signal Catalog chooser;
• static reporting keeps canonical model DataSet order as report-index authority;
• the DataSet directory captured by Smart Discovery is verification evidence only;
• count/order mismatch fails closed before RCB activation;
• no second DataSet-directory read is issued during static planning;
• live and SCL static modes use the same source-neutral configured-static activation API from engine PR Fix IO FAT live binding and relay fascia #137;
• projection authority precedence is consistently SCL model first, then live-discovery model;
• diagnostics now list exact unresolved primary DataSet memberships instead of only a count;
• dynamic DataSet writes and cyclic MMS process polling remain disabled in DataSet Signals mode.

Safety / acceptance

This PR is a candidate repair, not physical acceptance authority. PR #332 remains invalidated and closed.

Do not merge or publish until:

• engine PR Fix IO FAT live binding and relay fascia #137 CI is green;
• normal ARSAS build/test gates compile the promoted route without build-time patch dependence;
• physical discovery shows the Smart-Capture marker/timing and stays inside the frozen request budget;
• DataSet membership/order validation passes;
• configured RCB + GI produces correctly mapped InformationReport values without Boolean/quality cross-projection;
• cyclic MMS process polling remains zero;
• exact unresolved membership diagnostics are reviewed if runtime coverage is still below 58/58.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

[masarray]

Closing as superseded by the production Smart Discovery route that was promoted and physically accepted before v1.6.40, plus the current static DataSet/report-only guards on main. This branch is 124 commits behind current main and still references the old convergence authority path. Its original regression is addressed in the accepted production lineage; merging this branch now would risk reintroducing old engine/evidence/runtime assumptions. Keep this PR as historical diagnosis only.