fix(ci): run GPU E2E for merge groups

Signed-off-by: Evan Lezar <elezar@nvidia.com>

Author: elezar

.github/workflows/branch-e2e.yml

@@ -46,9 +46,12 @@ jobs:
               run_kubernetes_ha_e2e="$(jq -r 'index("test:e2e-kubernetes") != null' <<< "$LABELS_JSON")"
               ;;
             merge_group)
+              # Merge groups have no PR labels. If GPU E2E skips, its required gate
+              # stays pending until the queue times out and ejects the PR. HA stays
+              # off until stable.
               run_core_e2e=true
-              run_gpu_e2e="$(jq -r 'index("test:e2e-gpu") != null' <<< "$LABELS_JSON")"
-              run_kubernetes_ha_e2e="$(jq -r 'index("test:e2e-kubernetes") != null' <<< "$LABELS_JSON")"
+              run_gpu_e2e=true
+              run_kubernetes_ha_e2e=false
               ;;
             *)
               run_core_e2e=true