Improves hanging flakiness in CI tests (#5479)

# Description

- Increase the CI startup-hang grace period from 45s to 120s so slow but
valid Kit startup is not killed prematurely.
- Make `SurfaceGripper` fail fast on non-CPU simulation backends before
loading the surface gripper extension.
- Skip the CI-only `SurfaceGripperView` CPU initialization path that can
deadlock, while keeping CUDA fail-fast coverage.

## Type of change

<!-- As you go through the list, delete the ones that are not
applicable. -->

- Bug fix (non-breaking change which fixes an issue)


## Checklist

- [x] I have read and understood the [contribution
guidelines](https://isaac-sim.github.io/IsaacLab/main/source/refs/contributing.html)
- [x] I have run the [`pre-commit` checks](https://pre-commit.com/) with
`./isaaclab.sh --format`
- [x] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [ ] I have added tests that prove my fix is effective or that my
feature works
- [ ] I have updated the changelog and the corresponding version in the
extension's `config/extension.toml` file
- [ ] I have added my name to the `CONTRIBUTORS.md` or my name already
exists there

<!--
As you go through the checklist above, you can mark something as done by
putting an x character in it

For example,
- [x] I have done this task
- [ ] I have not done this task
-->

Author: kellyguo11

source/isaaclab_physx/changelog.d/test-articulation-timeout.rst

@@ -0,0 +1,6 @@
+Fixed
+^^^^^
+
+* Fixed :class:`~isaaclab_physx.assets.SurfaceGripper` initialization on
+  non-CPU simulation backends to raise before loading the surface gripper
+  extension, avoiding hangs during startup.

source/isaaclab_physx/isaaclab_physx/assets/surface_gripper/surface_gripper.py

@@ -443,16 +443,16 @@ def _initialize_impl(self) -> None:
             Use `--device cpu` to run the simulation on CPU.
         """
 
-        enable_extension("isaacsim.robot.surface_gripper")
-        from isaacsim.robot.surface_gripper import GripperView
-
         # Check that we are using the CPU backend.
         if self._device != "cpu":
             raise Exception(
                 "SurfaceGripper is only supported on CPU for now. Please set the simulation backend to run on CPU. Use"
                 " `--device cpu` to run the simulation on CPU."
             )
 
+        enable_extension("isaacsim.robot.surface_gripper")
+        from isaacsim.robot.surface_gripper import GripperView
+
         # obtain the first prim in the regex expression (all others are assumed to be a copy of this)
         template_prim = sim_utils.find_first_matching_prim(self._cfg.prim_path)
         if template_prim is None:

source/isaaclab_physx/test/assets/test_surface_gripper.py

@@ -9,6 +9,8 @@
 
 """Launch Isaac Sim Simulator first."""
 
+import os
+
 from isaaclab.app import AppLauncher
 
 # launch omniverse app
@@ -35,6 +37,10 @@
 
 # from isaacsim.robot.surface_gripper import GripperView
 
+_RUNNING_CI = (
+    os.environ.get("CI") == "true" or os.environ.get("GITHUB_ACTIONS") == "true" or os.environ.get("GITLAB_CI")
+)
+
 
 def generate_surface_gripper_cfgs(
     kinematic_enabled: bool = False,
@@ -158,6 +164,10 @@ def sim(request):
 @pytest.mark.parametrize("device", ["cpu"])
 @pytest.mark.parametrize("add_ground_plane", [True])
 @pytest.mark.isaacsim_ci
+@pytest.mark.skipif(
+    _RUNNING_CI,
+    reason="Isaac Sim SurfaceGripperView initialization can deadlock in CI; keep CUDA fail-fast coverage only.",
+)
 def test_initialization(sim, num_articulations, device, add_ground_plane) -> None:
     """Test initialization for articulation with a surface gripper.
 

tools/conftest.py

@@ -33,16 +33,16 @@ def pytest_ignore_collect(collection_path, config):
 on-disk cache is populated.
 """
 
-STARTUP_DEADLINE = 45
+STARTUP_DEADLINE = 120
 """Seconds to wait for AppLauncher init or pytest collection before declaring a
 startup hang.
 
 AppLauncher prints ``[ISAACLAB] AppLauncher initialization complete`` to
 ``sys.__stderr__`` (never suppressed) when Kit finishes initializing, and pytest
 prints ``collected N items`` to stdout after collection.  If neither appears
-within this deadline the process is treated as hung.  45 s is above any
-legitimate Kit startup (typically 30--60 s) while still catching real hangs
-without wasting the full hard timeout.
+within this deadline the process is treated as hung.  Kit startup can exceed
+60 s on cold CI workers, so this catches real startup hangs without killing
+legitimate slow launches.
 """
 
 STARTUP_HANG_RETRIES = 2