Set max parallel jobs on slow test runners (huggingface#7878)

* set max parallel

* update

* update

* update

Author: DN6

.github/workflows/push_tests.yml

@@ -56,6 +56,7 @@ jobs:
     needs: setup_torch_cuda_pipeline_matrix
     strategy:
       fail-fast: false
+      max-parallel: 8
       matrix:
         module: ${{ fromJson(needs.setup_torch_cuda_pipeline_matrix.outputs.pipeline_test_matrix) }}
     runs-on: [single-gpu, nvidia-gpu, t4, ci]