feat: improve app workflows and sdk diagnostics

[Pluze]

Why

Complete the current develop delivery as one coherent product update: repair DIC, file-selection, autosave, gait, curvature, diagnostic, busy-state, and launcher workflows; clarify SDK internal ownership; and make validation and PR preparation source-driven.

What changed

• Repaired DIC preprocessing interactions, crop/mask previews, point-only anchors, and robust rigid rotation/translation registration.
• Standardized validated multi-file, folder, and recursive source collection while retaining App-owned format meaning and alerts.
• Restored Video Marker point-change autosave, rebuilt Gait as a fitted 2-by-2 preview, and corrected Curvature insertion, overlay, and combined measurement behavior.
• Simplified complete-sensitive Session Log viewing/export, added exact and compact diagnostic state, automatic utility artifacts, delayed non-reentrant busy feedback, and complete-row selection.
• Layered App SDK internals by subsystem, split launcher/runtime workflows, and concentrated downstream App tests behind stable labkittest seams; production Apps contain no labkit.app.internal calls.
• Reordered Launcher Family/App columns, shortened Doc Generation, clarified targeted iteration/CI repair policy, and added a reusable squash-PR inventory with metadata-move-aware version checks.
• Consolidated the net change into six structured history records. All 18 versioned components advance exactly one direct step from the main baseline and each has one changed history record and owning manual.

Evidence

Integration policy: passed for origin/main d45e1553 -> develop d6a94428.
PR inventory: 331 changed paths; 18 direct component transitions; one changed history record and changed owning manual for every transition.
Skill validation: 7 repository Skill contracts passed; PR inventory script exercised on the real boundary; 8 integration-policy regression tests passed.
Authored links: 257 Markdown files checked, 0 unresolved or repairable links.
docsCheck: two independent trees matched, 384 generated files each.
changedFast (run once): selected 67 identities; 65 passed; 2 stale catalog expectation assertions failed.
Focused repair: TestCatalogSpec/parameterizedDefinitionsKeepEachPublicAppSelectable and TestCatalogSpec/definitionFileSelectsItsCompleteConformanceEvidence passed after updating the expected second definition contract. changedFast was not repeated.
CI repair evidence: the exact Gait selection identity passed 1/1; Session Log export plus Gait four-panel specifications passed 5/5; adjacent Gait and EIS workflows passed 3/3. The repair loop then allowed each hosted matrix to finish and batched the complete failure set before later pushes.
Required PR CI: run 30865406135 passed policy, documentation, Linux R2022b/all, Windows R2022b/all, Linux latest/all, Windows latest/core and hidden GUI, macOS latest/core and hidden GUI, path isolation, and aggregate CI Gate.
Private downstream compatibility: complete private suite passed 50/50 (48 headless, 2 hidden-GUI) through the new labkittest seams.

Risks and follow-up

Required PR CI established the complete Linux, Windows, and macOS headless/hidden-GUI/path-isolated claim. Remaining developer-led checks are native file-dialog feel, DIC pointer behavior and representative-image registration suitability, dense real-recording Gait readability, Curvature pointer feel, and delayed busy-state visual timing. Compact diagnostic state preserves structure but intentionally replaces individually large supported values and is not scientific input. No private App details or supplied lab artifacts are tracked.

Author confirmation

• The final diff matches the stated scope and contains no unrelated change.
• The evidence above covers required local and manual validation; every remaining unverified boundary is explicit.
• The diff contains only synthetic or generic public data, with no private App details, lab identifiers, local paths, or recognizable sample values.