fix(cua): move sessions_reload before typescript; fix CI emulator boot

sessions_reload regression phase now runs BEFORE the TypeScript task.
Previously it was gated behind typescript which fails in CI when the model
is unavailable — meaning the actual sessions regression check never ran.

Phase order is now:
  connect → session_list (pre-created session required) → new_session
  → sessions_reload (navigate back, list must be non-empty) [CRITICAL]
  → typescript (informational) → verify (informational) → settings

Critical phases: connect, session_list, new_session, sessions_reload.
TypeScript/verify/settings are informational (model availability varies).

CI emulator fixes:
- api-level: 30 → 28 (more stable, boots reliably on ubuntu-latest)
- target: google_apis → default (lighter, no Play Services needed for
  sessions regression test, avoids known boot issues with google_apis)
- disable-animations: true (reduces boot overhead)
- emulator-boot-timeout: 600 (explicit, matches action default)
- Switch from --scenarios to --showcase (runs the new structured flow
  with _precreate_test_session + sessions_reload phase)
- Clear app state before install (pm clear) for deterministic first-run

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

.github/workflows/cua-smoke.yml

@@ -238,30 +238,26 @@ jobs:
       - name: Run CUA smoke test with emulator
         uses: reactivecircus/android-emulator-runner@v2
         env:
-          MODEL_CAPABLE: ${{ steps.probe.outputs.MODEL_CAPABLE }}
-          SCENARIOS: ${{ steps.probe.outputs.SCENARIOS }}
+          OPENCODE_URL: ${{ env.OPENCODE_URL }}
         with:
-          api-level: 30
+          api-level: 28
           arch: x86_64
-          target: google_apis
+          target: default
+          disable-animations: true
+          emulator-boot-timeout: 600
           emulator-options: -no-window -no-audio -no-boot-anim -gpu swiftshader_indirect -no-snapshot
           # NOTE: android-emulator-runner runs this script with /usr/bin/sh (dash).
-          # Avoid bash-only constructs like multi-line `|| { ... }` brace groups.
           script: |
-            # Non-fatal re-check; server health was already gated in the prior step.
-            curl -sf http://127.0.0.1:4096/global/health || echo "WARN: opencode health re-check failed (started in prior step)"
+            # Clear any stale app state so the showcase starts with a fresh connection screen.
+            adb shell pm clear cc.agentlabs.opencode || true
             adb install android/app/build/outputs/apk/release/app-release.apk
             adb shell am start -n cc.agentlabs.opencode/.MainActivity
             sleep 5
-            # SCENARIOS is computed in the probe step (bash) and passed in via env, so
-            # this script stays single-line — dash (used by android-emulator-runner)
-            # mangles multi-line if/fi blocks. The widened core journey is
-            # connect -> create session -> code -> verify -> list; coding_task is only
-            # included when the model probe confirmed opencode can reply.
-            echo "Running scenarios: ${SCENARIOS}"
-            # --max-steps raised so each scenario (especially coding_task with its
-            # long wait-for-completion loop) has enough budget.
-            python3 scripts/android-cua-smoke.py --model gpt-5.4 --include-xml --max-steps 60 --scenarios "${SCENARIOS}"
+            # --showcase runs the structured onboarding flow which includes:
+            #   session_list: pre-created session MUST appear (not empty-list-ok)
+            #   sessions_reload: navigate back from session, list must be non-empty
+            # These are the regression guards for the "sessions not loading" bug.
+            python3 scripts/android-cua-smoke.py --showcase --model gpt-5.4 --include-xml --opencode-url "${OPENCODE_URL}"
 
       - name: opencode server log
         if: always()