01a07197 - Stabilize E2E readiness and fix nullable KYC fixture inputs

[TaprootFreakAI]

EN:
KYC and financial-dashboard E2E tests now wait for completed responses and the rendered result before checking access or navigating again. Regression tests hold real API responses to expose premature checks and navigation; nullable KYC fixture inputs also now satisfy the harness type check.

DE:
Die E2E-Tests für KYC und das Finanz-Dashboard warten jetzt auf abgeschlossene Antworten und den dargestellten Endzustand, bevor sie Zugriffe prüfen oder weiternavigieren. Regressionstests halten echte API-Antworten zurück, um zu frühe Prüfungen und Seitenwechsel sichtbar zu machen; nullable KYC-Fixture-Eingaben erfüllen zudem den Typcheck.

Implementation and validation

KYC absence assertions could run before the file response arrived, turning an existing expected failure into an unexpected pass. The financial hub test could leave a destination while its requests were still running, producing aborted requests and console errors.

Wait for the KYC response and terminal screen before checking the intended result, and for financial response bodies and destination UI before navigating again. The regressions hold delivery of real API responses and have a matching reality declaration. Removing the two wait barriers makes both regressions fail at the premature-completion assertions.

The KYC race fixture normalizes nullable optional arguments to undefined; its factory already converts either nullish value to SQL null, preserving runtime behavior. GitHub found this pre-existing TS2322 in the separate harness type check before E2E execution.

Validation of the initial synchronization commit a35b632 (completed before the requester selected GitHub-only App CI):

• Node 20 Linux: dependency installation, lint, Markdown check, development app build and development widget build passed.
• Focused real-stack run: 7 ordinary passes and 1 existing expected product failure.
• Mutation control: both regressions reject removed wait barriers; 4 setup tests pass.
• Full fresh-stack run: 264 cases, comprising 256 ordinary passes, 5 existing expected failures and 3 existing skips, in 8.4 minutes; the route-coverage gate passed.
• Final head f39381fb: full GitHub PR CI passed on GitHub-hosted runners with ci:full: 1,828 unit tests in 115 suites; lint, Markdown, both builds and harness type check passed; full E2E ran 264 cases (256 ordinary passes, 5 existing expected failures, 3 existing skips, 0 unexpected failures, 0 flaky outcomes) in 10.7 minutes, including the route-coverage gate.
• Handbook build and container smoke, CodeQL and the PR workflow checks passed at the same head. The CI-label helper correctly skipped because ci:full was already present; the earlier Draft-only Handbook skip was superseded by the successful Ready-triggered run.

The change is limited to the E2E harness and its documentation. No Jest-instrumented production file or visual flow changes. Existing expected product failures and skipped coverage remain explicitly tracked, including #1289 and #1290; their resolution is not claimed by this synchronization fix.

Independent final reviews were explicitly waived by the requester: 0 independent final-review approvals are claimed. Downstream consumers of this harness need the fix on develop before their next validation uses the corrected tests.

[TaprootFreakAI]

EN:
Independent final reviews were explicitly waived by the requester, so 0 independent review approvals are claimed.
Wait for completed KYC and financial screens and normalize nullable KYC fixture inputs without changing stored values.

DE:
Die unabhängigen Abschlussreviews wurden ausdrücklich erlassen, deshalb werden 0 unabhängige Review-Freigaben beansprucht.
KYC- und Finanztests warten auf fertige Bildschirme und normalisieren nullable KYC-Fixture-Eingaben ohne Änderung der gespeicherten Werte.

Validation and final gates

Final head: f39381fb3a2f9dfbfa952a860f2d66f3d71bbfc4; both commits have verified signatures.

Full GitHub PR CI passed on hosted runners: 1,828 unit tests in 115 suites, lint, Markdown, app/widget builds, harness type check and full E2E. The browser report records 264 cases: 256 ordinary passes, 5 existing expected failures, 3 existing skips, 0 unexpected failures and 0 flaky outcomes; duration 10.7 minutes; route-coverage gate passed.

Handbook build and smoke, both CodeQL analysis jobs and the PR check passed at this head. The CI-label helper skipped correctly because ci:full was already set; the prior Draft Handbook skip was superseded by the successful Ready-triggered run. No local App CI was run for this final head, as explicitly requested.

All four comment sources were read with pagination: no issue comments, reviews, inline comments or review threads required action. Mergeability is MERGEABLE, the branch includes current develop, and the final title/body describe all seven changed files. Branch protection/rules do not add required status checks. The independent final-review waiver is recorded explicitly and does not substitute for these CI results.

The earlier synchronization-only snapshot was additionally checked with held real responses and mutation controls: removing the waits caused both regression assertions to fail. Existing expected product failures and skipped cases are retained explicitly; they are not claimed as fixed.