Add E2E test retry jobs for PRs to handle flakiness

On PRs, E2E tests (iOS/Android, RNTester/TemplateApp) now retry up to
2 additional times on failure. Each retry runs on a fresh runner to
address environment-level flakiness.

- Original E2E jobs use `continue-on-error` on PRs so failures don't
  block the workflow
- Step-level outcome is captured as a job output to trigger retries
- Retry jobs only run on `pull_request` events
- On `main`, behavior is unchanged: `continue-on-error` is false and
  the existing `rerun-failed-jobs` mechanism handles retries
- Added `overwrite: true` to artifact uploads in maestro composite
  actions so retry jobs don't fail on duplicate artifact names

Author: cipolleschi

.github/actions/maestro-android/action.yml

@@ -77,6 +77,7 @@ runs:
       if: always()
       with:
         name: e2e_android_${{ steps.normalize-app-id.outputs.app-id }}_report_${{ inputs.flavor }}_${{ inputs.emulator-arch }}_NewArch
+        overwrite: true
         path: |
           report.xml
           screen.mp4
@@ -85,4 +86,5 @@ runs:
       uses: actions/upload-artifact@v6
       with:
         name: maestro-logs-android-${{ steps.normalize-app-id.outputs.app-id }}-${{ inputs.flavor }}-${{ inputs.emulator-arch }}-NewArch
+        overwrite: true
         path: /tmp/MaestroLogs

.github/actions/maestro-ios/action.yml

@@ -69,6 +69,7 @@ runs:
       uses: actions/upload-artifact@v6
       with:
         name: e2e_ios_${{ inputs.app-id }}_report_${{ inputs.flavor }}_NewArch
+        overwrite: true
         path: |
           video_record_1.mov
           video_record_2.mov
@@ -81,4 +82,5 @@ runs:
       uses: actions/upload-artifact@v6
       with:
         name: maestro-logs-${{ inputs.app-id }}-${{ inputs.flavor }}-NewArch
+        overwrite: true
         path: /tmp/MaestroLogs