Complete SV Subscriber P3C-P3F evidence pipeline

[masarray]

What changed

P3C — SCL expected configuration

• Convert safely bound SCL publisher profiles into SvExpectedStreamConfiguration.
• Run configured-versus-observed comparison in the shared live/PCAP observation pipeline.
• Use Compatible by default and retain explicit Strict mode.
• Persist expected configuration, field findings, and exact/warning/error summaries.
• Reject unsafe or ambiguous SCL candidates before applying their dataset layouts.

P3D — Compact Subscriber UX

• Add compact PROFILE, CONFIDENCE, SCL MATCH, and WINDOW state.
• Add collapsed evidence details for profile, configuration, and observation findings.
• Preserve Compatible warnings without forcing BAD; Strict errors remain blocking.
• Prevent rejected SCL bindings from contaminating payload decoding.
• Use a two-second high-rate observation window and stable complete two-cycle visuals.
• Replace per-item collection churn with bulk refreshes to reduce flicker.

P3E — Evidence reports

• Add versioned evidence schema arsvin.sv-subscriber-evidence/v1.
• Export paired Markdown and JSON from the same runtime/observation snapshot.
• Serialize capture metadata, stream identity, runtime integrity, window facts, counter transitions, provenance, profile evidence, expected SCL configuration, comparison findings, diagnostics, phasors, software version, and commit.
• Validate schema, generation time, product identity, stream count, and unique stream keys.

P3F — Baseline-versus-candidate comparison

• Add versioned comparison schema arsvin.sv-subscriber-evidence-comparison/v1.
• Add a Subscriber Compare action that accepts baseline and candidate evidence JSON files and exports paired Markdown and JSON comparison bundles.
• Match exact stable stream keys first, then use logical identity (APPID, destination MAC, VLAN, svID, dataset) so source-MAC/NIC failover is reported as a change rather than false removed/added streams.
• Classify added, removed, changed, and unchanged streams.
• Rank changes as Info, Warning, or Error across health, source MAC, confRev, ASDU packing, sampling fields, runtime counters, measured rates, observation windows, SCL binding, configuration findings, profile confidence, dataset signature, provenance, and diagnostics.
• Treat removed streams, BAD/ERROR health regression, out-of-order/payload failures, blocking configuration errors, profile conflicts, and dataset-signature changes as high-signal regressions.
• Suppress false rate regressions inside a one-percent tolerance.
• Keep comparison keys unique even when multiple publishers share one logical stream identity.

Deterministic coverage

• 5 P3C expected-configuration integration tests.
• 3 P3D/window presentation-contract tests.
• 3 P3E report schema/round-trip/validation tests.
• 4 P3F tests covering source failover matching, added/removed classification, Markdown/JSON round-trip, severity preservation, and rate tolerance.

Validation

The final clean head is running:

• locked NuGet restore;
• Publisher and Subscriber warnings-as-errors builds;
• full tests and coverage gates;
• documentation/site validation;
• licensing, provenance, and terminology validation;
• dependency vulnerability reports;
• CodeQL;
• portable EXE and Windows installer packaging/smoke tests.

The PR will be merged only after all required final-head workflows pass.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ff2282a8f4

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".