P0: adaptive multivendor SV measurement foundation

[masarray]

Scope

This draft starts the real-device/multivendor Subscriber hardening program using SMU615-style installed-base traffic as the first reference case, without adding vendor-name conditionals to the protocol core.

Implemented in this tranche

• provenance-aware raw-to-engineering scaling with conservative 9-2LE-style structural gates
• explicit raw-count fallback when units are not proven
• 50/60 Hz-neutral timebase resolution (no hidden 50 Hz assumption)
• robust smpCnt transition tracking for normal wrap, gaps, duplicates, out-of-order traffic, and trusted restarts
• rolling current stream health separated from cumulative session issue totals
• waveform/RMS/phasor processing on engineering values when scaling evidence is sufficient
• raw, scaled value, scaling confidence, timebase source, and reason visibility in the UI
• dynamic A/V/count waveform units and stable near-zero display scaling
• header clipping and incorrect green BAD status styling fixes
• deterministic engine tests for scaling, timebase, and counter behavior

Validation at head 31ded4f

• CI: passed
• Publisher warnings-as-errors build: passed
• Subscriber warnings-as-errors build: passed
• test and coverage gates: passed
• dependency vulnerability checks: passed
• Windows release build: passed
• CodeQL: passed

Safety boundary

Named vendor identity is not used as a scaling rule. Unknown or conflicting streams remain visible in raw counts. This is an engineering analyzer enhancement, not a conformance or calibrated-test claim.

Still pending after this draft tranche

• sanitized SMU615 SCL/PCAP golden evidence pack
• primary/secondary conversion from explicit CT/VT context
• detailed quality-bit semantic classification
• local Npcap drop/queue-overrun attribution
• IEC 61869-9 profile manifests and additional vendor fixtures

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 31ded4fe9d

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]

Keep SCL binding from certifying a scale

For any SCL-bound stream whose 64-byte 4I+4U payload uses device-specific raw multipliers, this condition accepts the stream without sampling evidence and the resolver then applies the fixed 0.001/0.01 conversions while labelling them SclBacked. In this codebase the SCL payload layout supplies element types and offsets, not a scaling multiplier, so a valid but differently scaled bound stream will display incorrect A/V values as SCL-backed engineering measurements. Require explicit scale/CT/VT context (or keep this path raw/provisional) rather than treating binding alone as scale evidence.

AGENTS.md reference: AGENTS.md:L24-L28

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Discard the old timebase after an unresolved reconfiguration

When an already resolved stream changes confRev, sample rate, or layout and the new evidence cannot resolve a timebase, this guard retains the prior configuration's frequency, samples-per-cycle, and counter wrap indefinitely. The subsequent loop uses that stale wrap to classify smpCnt, and the snapshot continues generating a waveform/phasors using the old cycle length, so a reconfigured live stream can be shown as healthy engineering data or spuriously report gaps rather than reverting to the intended unresolved-timebase state. Reset/replace the cached resolution when the configuration changes or its sampling declarations no longer match.

AGENTS.md reference: AGENTS.md:L7-L7

Useful? React with 👍 / 👎.