Move test-executorch-cuda-build to linux_job_v3 (pytorch#22890)

Drops the `driver-version` / `driver-download-url` pins rather than
porting them, because the job was mostly not using them:

```
12.6   minimum driver 560.28.03   R580 clears it
13.0   minimum driver 580.65.06   R580 is exactly this
13.4   asked for 615.71.09        see below
```

Two of the three cells were already passing `580.65.06`, which is what
OSDC nodes ship from the EKS NVIDIA AMI — those installs were a no-op.
Only 13.4 wanted something newer, and CUDA minor version compatibility
means a 13.x application runs on any driver supporting 13.0, provided it
stays off newer driver APIs. So R580 is likely enough there too.

Installing a driver per job is an EC2 idea regardless: on OSDC the
driver belongs to the node and a pod cannot replace it.

If 13.4 does fail, the fix is the R615 AMI in pytorch/ci-infra#1068, not
this workflow. The failure would look like a test error rather than a
build one — `test-cuda-build.sh` runs a CUDA matmul and exits 1 if the
device is unavailable.

### Testing


https://github.com/pytorch/executorch/actions/runs/35148091997/job/104969392177?pr=22890

Author: huydhn

.ci/scripts/tests/test_cuda_workflow.py

@@ -39,24 +39,17 @@ def test_build_matrix_preserves_existing_cuda_versions(self):
         )
         self.assertEqual(job["with"]["gpu-arch-version"], "${{ matrix.cuda-version }}")
 
-    def test_cuda134_driver_uses_matching_workflow_and_action_revision(self):
+    def test_cuda_builds_take_the_node_driver_on_an_unpinned_v3(self):
+        # This job used to pin linux_job_v2 to a test-infra SHA because that was
+        # the only ref carrying driver-version / driver-download-url, and it
+        # installed R615 for the 13.4 cell. On OSDC the driver belongs to the
+        # node and a pod cannot replace it, so the job takes what the node has.
         job = WORKFLOW["jobs"]["test-cuda-builds"]
-        workflow, revision = job["uses"].split("@")
         self.assertEqual(
-            workflow, "pytorch/test-infra/.github/workflows/linux_job_v2.yml"
-        )
-        self.assertRegex(revision, r"^[0-9a-f]{40}$")
-        self.assertEqual(job["with"]["test-infra-ref"], revision)
-        self.assertEqual(
-            job["with"]["driver-version"],
-            "${{ matrix.cuda-version == '13.4' && '615.71.09' || '580.65.06' }}",
-        )
-        self.assertEqual(
-            job["with"]["driver-download-url"],
-            "${{ matrix.cuda-version == '13.4' && "
-            "'https://download.nvidia.com/XFree86/Linux-x86_64/615.71.09/"
-            "NVIDIA-Linux-x86_64-615.71.09.run' || '' }}",
+            job["uses"], "pytorch/test-infra/.github/workflows/linux_job_v3.yml@main"
         )
+        for pin in ("test-infra-ref", "driver-version", "driver-download-url"):
+            self.assertNotIn(pin, job["with"])
 
     def test_cuda134_runtime_update_precedes_build_and_propagates_failure(self):
         script = WORKFLOW["jobs"]["test-cuda-builds"]["with"]["script"]

.github/workflows/cuda.yml

@@ -67,19 +67,15 @@ jobs:
         cuda-version: ["12.6", "13.0", "13.4"]
 
     name: test-executorch-cuda-build-${{ matrix.cuda-version }}
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@c3c4c4d48e97dbaaabd9b65496afd8c8d8dad713
+    uses: pytorch/test-infra/.github/workflows/linux_job_v3.yml@main
     permissions:
       id-token: write
       contents: read
     with:
       timeout: 90
-      runner: linux.g5.4xlarge.nvidia.gpu
+      runner: mt-l-x86aavx2-11-41-a10g
       gpu-arch-type: cuda
       gpu-arch-version: ${{ matrix.cuda-version }}
-      test-infra-ref: c3c4c4d48e97dbaaabd9b65496afd8c8d8dad713
-      driver-version: ${{ matrix.cuda-version == '13.4' && '615.71.09' || '580.65.06' }}
-      driver-download-url: ${{ matrix.cuda-version == '13.4' && 'https://download.nvidia.com/XFree86/Linux-x86_64/615.71.09/NVIDIA-Linux-x86_64-615.71.09.run' || '' }}
-      use-custom-docker-registry: false
       submodules: recursive
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |