fix(ci): capture test failure details before pod cleanup

The test-gate pod gets cleaned up immediately on failure, losing all
stdout. Pipe cargo test through tee and grep for FAILED/panicked
output before exiting, so failure logs are preserved in Argo.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

infra/k8s/argo/ci-pipeline-template.yaml

@@ -314,7 +314,13 @@ spec:
             cargo clippy --workspace --lib -- -D warnings
 
             echo "=== Running tests (lib only, no integration) ==="
-            cargo test --workspace --lib
+            cargo test --workspace --lib 2>&1 | tee /tmp/test-output.log
+            TEST_EXIT=${PIPESTATUS[0]}
+            if [ "$TEST_EXIT" -ne 0 ]; then
+              echo "=== TEST FAILURES ==="
+              grep -A5 'FAILED\|panicked\|test result: FAILED' /tmp/test-output.log || true
+              exit "$TEST_EXIT"
+            fi
 
             echo "=== Test gate passed ==="