Continue benchmark other models even if one fails to export

Author: huydhn

.github/workflows/android-perf.yml

@@ -205,6 +205,7 @@ jobs:
 
   # Let's see how expensive this job is, we might want to tone it down by running it periodically
   benchmark-on-device:
+    if: always()
     permissions:
       id-token: write
       contents: read

.github/workflows/apple-perf.yml

@@ -279,6 +279,7 @@ jobs:
           path: ${{ runner.temp }}/artifacts/
 
   benchmark-on-device:
+    if: always()
     needs:
       - set-parameters
       - upload-benchmark-app