fix(sv): truthful per-stream engineering scales, no-SCL 4I4V profile, safe power

[masarray]

Closes #26 after independent Windows candidate acceptance.

Physical evidence / root cause

ESP32-P4 produced valid 4I+4V INT32 + quality payload and 4000 fps; raw receiver plotted dynamic wave/phasor without SCL. Metering incorrectly labelled raw ~1000 current counts as ~1000 A, ~5774 voltage counts as 5.77 kV, and therefore invented MW/Mvar.

Correct engineering boundary

• Raw counts by default: channel samples, waveform, phasor and RMS remain available, accurately labelled counts. Physical P/Q/S/PF show N/A until operator explicitly chooses a scaling authority. No auto-inference from svID, APPID, vendor or payload shape.
• One-click nominal 4I+4V profile per strong stream key: explicitly converts I=0.001 A/count and U=0.01 V/count; no SCL required for bench profile, but UI warns this is operator-selected and not calibrated/proven by packets. Changes for one stream never apply to another.
• All values are derived from existing selected SvStreamState, raw samples and coherent windows. Scale multiplication only on UI snapshot projections, not per-packet hot path. No second decoder/cache. Consistent waveform series units and RMS/phasor units; no inconsistent MW on raw data.
• Copy Evidence includes scale authority, selected snapshot RMS, watts/vars, provenance and packet diagnostics.
• Deterministic tests of raw default, scaling/reversal without changing phase and separation of two simultaneous stream identities.

Guardrails

No capture or publisher commands, no customers' PCAP committed, no auto-universal IEC physical scaling, no hardware timing claims. Existing raw waveform, SCL matching, timing and receive-only architecture protected. Await repo health / dotnet solution test / Windows candidate CI before merge. Further bench tests with operator PCAP required for physical acceptance.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-16T03:49:55.728251Z	0768e86	PR opened

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 0768e8644b

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".

[masarray]

Physical acceptance handoff — merged 2026-10-16

• PR fix(sv): truthful per-stream engineering scales, no-SCL 4I4V profile, safe power #27 merged into main at 07364f550e74797e4a1a35f8a1f517b9a27a69cf, merge tree 3e9068550868a9173763b8e8a82a61061b1e7159 exactly equals tested head 9fd74d6073a7bfeb7b0298e09dd17466e91a6db7.
• Exact reviewed HEAD: CI / repository health success, test suite 59/59 pass (candidate TRX); Runtime multi-stream success; CodeQL success; dependency checks success; Windows candidate packaging and verification success. Both inline P1 review threads resolved.
• Candidate Actions workflow: https://github.com/masarray/DigSubAnalyzer/actions/runs/38847933759 · artifact 12023717572 ProcessBusInsight-v1.4.0-beta.2-win-x64-candidate-62. Outer artifact SHA256 4509a2b45bf357fb51c1a6fa474484c29fee5b4d8b7d64f392e1e45e4f384dee. Inside portable ZIP SHA256 61f6cea4ab0cf49a0ee98548eb10dfef28e65b9e3ae92e84c4341d4cada6c8c1. Manifest source commit 9fd74d6073a7bfeb7b0298e09dd17466e91a6db7, executable ProcessBusInsight.exe present. Engineering documentation docs/SV_ENGINEERING_SCALE_LAB.md.
• Operator lab recipe: unzip candidate (no install; run ProcessBusInsight.exe), select ASIX NIC, Run capture; choose canonical ARSTACK_SV01, check waveform+phasor. Metering's default Raw counts (uncalibrated) labels RMS counts and gates P/Q/S/PF. Switch SV units explicitly to 4I+4V nominal (1 mA / 10 mV per count) for verified lab profile: expected ≈1 A, 57.74 V, 50 Hz, 4000 fps, phasor responds to live angle changes, P/Q/S finite only after physical scale chosen. Separate AR_DIAG_SV1 must stay raw until independently selected. Metering has Copy diagnostics button for text evidence.
• Scope: raw receive-only; no customer PCAP committed; physical test using this new candidate remains for user. Windows capture timestamps cannot validate PTP sync accuracy. No certified calibration or IED subscriber acceptance claimed.
• ARStack Studio firmware-update success has been confirmed by operator on separate source; no firmware changes here. Issue P0 SV metering: raw INT32 counts must not appear as calibrated A/V or MW without scale authority #26 auto-closed by merge; create follow-on issue for any independent physical findings rather than reverting nominal scale safety.