FAT v2 P5: persist, package and report Value 1/Value 2 evidence

[masarray]

P5 — GO

Stacked on exact proven P4 head 8e548356ad2269a1f2636feb88b0cb4213a82e28.

Final exact proven P5 head:
45eb6bf31755b6d350d70d5c973140a5adaad5a0

This phase closes FAT v2 persistence/package/report compatibility without collapsing the SCL-driven domain into the legacy ON/OFF IoTestProject model.

Contract delivered

• persist operator disposition plus current Value 1 / Value 2 evidence by immutable static-membership SignalId
• same-SCL reopen recreates authoritative DataSet rows first, then restores saved operator state/evidence
• snapshot source-set mismatch and signal-set mismatch fail closed
• dedicated SCL-backed FAT v2 .arsas package kind; no fake workbook
• package contains immutable SCL source bundle, FAT v2 snapshot, PDF report, XLSX report, and README
• package reopen verifies manifest/source hashes, recreates engineering rows through ARIEC SclWorkspaceService, then restores snapshot
• packaged source extraction preserves the original source filename while isolating files by SourceId so the P3 filename-sensitive source identity remains stable
• generic FAT v2 PDF/XLSX report fields use Value 1 / Value 2 and include removed rows/disposition
• FAT launcher can open raw SCL, reopen a FAT v2 .arsas, and save the active FAT v2 workspace as a portable .arsas
• workspace close writes the durable local snapshot
• legacy IO FAT package service remains untouched
• no injected/reference value, tolerance, error %, or analog auto-PASS logic
• no ARIEC engine-pin change

Regression coverage

• same SCL reopen preserves ExcludedByOperator + V1/V2 evidence
• portable .arsas round-trip preserves source identity, disposition, V1/V2 and remains SCL-backed
• tampered packaged SCL is rejected before engineering state is trusted
• PDF/XLSX use generic Value 1 / Value 2 terminology and preserve removed rows

Scope audit

Compare P4 exact -> P5 exact:

• ahead_by: 9
• behind_by: 0
• 7 changed files
• no ARIEC lock change
• no protocol/control changes
• no new SCL/XML parser
• legacy workbook FAT package implementation not rewritten

Changed files:

1. GlobalUsings.FatV2.cs
2. MainWindow.FatV2.cs
3. Services/IoTesting/FatSclWorkspaceBootstrapService.cs
4. Services/IoTesting/FatVerificationPackageService.cs
5. Services/IoTesting/FatVerificationPersistenceService.cs
6. Services/IoTesting/FatVerificationReportService.cs
7. tests/ARSAS.Tests/FatVerificationPersistencePackageTests.cs

Exact-head CI evidence

All required workflows completed successfully on exact head 45eb6bf31755b6d350d70d5c973140a5adaad5a0:

1. Validate IO List Testing #551 — run 33576455333 — SUCCESS

   • focused restore SUCCESS
   • focused IO testing build SUCCESS
   • IO testing regression suite SUCCESS

2. Validate SV evidence bundles #717 — run 33576455324 — SUCCESS

   • focused compiler SUCCESS
   • deterministic SV evidence regression suite SUCCESS

3. Build ARSAS #1608 — run 33576455327 — SUCCESS

   • source/website/license boundaries SUCCESS
   • premium UX/GOOSE/SMV/SAS/release invariants SUCCESS
   • immutable ARIEC API verification SUCCESS
   • solution restore/build SUCCESS
   • full ARSAS application regression suite SUCCESS
   • portable single EXE publish SUCCESS
   • portable single EXE smoke test SUCCESS
   • portable artifact upload SUCCESS

Earlier intermediate P5 heads exposed compile-only integration defects (missing IoTesting model import and a Brush name collision). Those were corrected before this exact-head qualification. No runtime/regression failure is hidden by the final evidence.

P5 = GO.

PR remains draft and unmerged.

[masarray]

Closing as superseded by consolidated FAT PR #236. This divergent P5 release-proof line is not the active authority anymore; persistence/package/report behavior has since been integrated and evolved on the consolidated workspace branch. Branch/commits remain preserved for historical comparison.