P0: restore active Browser panel routing and enforce rendered QA

[masarray]

Recover from the blank/misaligned workbench regression. The previous hotfix 009855c addressed shadowed QML services, but this UX branch still retained the obsolete StackLayout/Repeater slot mapping despite validated routing changes already merged on main.

• Port the validated persistent plain-Item Browser host and explicit active-slot visibility from the accepted main baseline.
• Keep root.browserFleet service bindings for each independent IED.
• Restore validated C++ rendered --qa-browser-routing smoke/fixture path; enforce it in Qt and Release Hardening CI.
• Update P6B/P6E/delegate contracts to match the actual non-shadowed, accurately routed shell.
• No MMS, SCL or report wire-runtime changes.

Additional root cause confirmed by rendered CI

• MmsClientWorkspace.qml and IedBrowserNavigation.qml declared required property int index but used model.* without required property var model. Qt 6 does not inject the implicit model context when any required delegate properties are declared. CI reproduced ReferenceError: model is not defined, blank delegate values, and a missing signal-tree QA identity.
• Explicitly bind required property var model, restore observable tree/value-table object names, and require the first rendered model value label to equal the fixture IED name. QA now also checks that all per-IED controller pointers match the active slot.
• Earlier CI red is retained as failure evidence. Exact-head validation and real-device behavior remain outstanding.

Do not merge or publish a portable build until exact-head CI is green and rendered/fixture checks pass. main remains unchanged.

Formal acceptance milestones

• R0 GUI recovery and exact-head release gate: Milestone R0 — Restore usable IEC 61850 Browser before UX expansion #128.
• R1 canonical Open SCL / Discovery → static reporting: Milestone R1 — Open SCL and Discovery converge into verified static reporting #129 (begins only after R0 evidence).

The latest CI found a real failure (unbound rendered label) rather than a compilation issue. The rendered QA now walks QQuickItem visual children. File/Home self-binding and transient DataSet ownership dereference are corrected. Qt/Release gate now rejects QML ReferenceError, TypeError, and binding loops. Keep Draft until exact-head checks are green and the real IED test is documented.

[masarray]

R0 diagnostic / lifecycle fix (2026-09-27)

Last tested head 48728708 had Qt and Linux Release Hardening failure, even though all protocol-side CI passed and the Windows package job succeeded independently.
The rendered fixture actually printed BROWSER_FLEET_ROUTING_PASS (correct active tab, offline signal and Model Values text, reindex) but then emitted many TypeError: ... of null as per-IED service QObjects were destroyed while Repeater views still held live bindings. These errors were correctly caught by the new log gate; they are not evidence of missing model data.
Changes in this follow-up: defer retirement of a closed slot through the next Qt event-loop turn; detach Browser Repeater delegates in aboutToQuit before fleet C++ destruction; retain source regression contracts; enforce Linux rendered acceptance as a prerequisite for publishing the Windows RC artifact.
Exact-head validation is pending. Do not merge or link stale artifacts as a release. Milestone #128 tracks R0 exit; #129 tracks SCL/Discovery static report parity.

[masarray]

R0 exact-head validation update — 8b077f5e

• IED Simulator Qt: success. The rendered active-slot/offline signal/value label test and subsequent P6A/P6B/P6C workflow contracts succeeded.
• Release Hardening: Linux release-soak success (rendered Browser, QML runtime diagnostic gate, MMS reconnect, app-close child/socket soak); gated Windows package is still building.
• Source cause fixed: services are now frozen to the corresponding Browser delegate at creation; removed slot lifetime is retained briefly for visual teardown; delegates detach before application shutdown. This prevents QML service pointers from transiently re-binding to the wrong IED or becoming null.
• Remaining R0 gates: full Windows package success + exact-head artifact validation, offline Open SCL GUI walkthrough, real AA1E1F06R4 device acceptance, and deliberate promotion strategy without overwriting main.

[masarray]

R0.1 compact Browser command surface (#135)

Extended recovery branch at 3a3ceef9 with a two-row, 96px model/connection command bar. The former single row could overflow the supported 1024px width, causing apparently blank or inaccessible commands. Top row: model source/Save SCL/identity. Bottom row: IP, TCP port, context-aware Online/Discover/Disconnect, and session status. Status uses compact ellipsis and full tooltip.
Added stable object identities for command bar, host/port and status, plus rendered Qt QA at 1024px asserting their actual visual geometry while retaining the existing nonblank signal tree/Model Values and multi-IED routing checks. Regression contract rejects reintroduction of the 68px single-row header.
All code is pushed; exact-head CI is running. This is not complete parity with IEDScout or real-device acceptance; next exit checks are #135 and #129. R0 previously green on 8b077f5e remains the recovery baseline, and main remains untouched.