fix(distributed): subgroup NCCL regression + nccl_utils TRT 11 cleanup

[apbose]

test_native_nccl.py:

• Fix test_distributed_mode_subgroup and test_distributed_mode_context_switch regressions introduced by PR Python runtime rework  #4222 (Python runtime rework): add dist.barrier(group=subgroup) before first TRT forward so PyTorch's lazy ncclComm_t is initialized and bind_nccl_comm() sees a non-zero getCommPtr().
• Remove all use_python_runtime=True flags (deprecated no-op in 2.13).
• Add _cpp_runtime_available() guard to TestPythonRuntimePickle so it skips in C++ builds where _TRTEngine's custom op is never registered.
• Fix _multirank_scatter_correctness / _multirank_gather_correctness in run_multirank_tests(): wrap with lambdas that loop over root ranks to match their (root, rank, world_size, device) signature (broken since PR TRT 11 MD Ops #4321).
• Remove _multirank_distributed_mode_subgroup_python and _multirank_distributed_mode_context_switch_python (cannot work when C++ extension is loaded); collapse _multirank_pg_migration to cpp-only.

_nccl_utils.py:

• setup_nccl_for_torch_tensorrt(): drop ensure_nccl_symlink() call and pre-load libnccl.so.2 directly instead of via the unversioned symlink. TRT 11.0+ (ncclWrapper.cpp) tries dlopen("libnccl.so.2") first, matching PyTorch's soname exactly; glibc reuses the already-loaded handle so no symlink is needed. ensure_nccl_symlink() is kept for TRT < 11.0 users.
• Update module docstring to reflect confirmed TRT 11.0+ behavior.

[github-actions]

There are some changes that do not conform to Python style guidelines:

--- /home/runner/work/TensorRT/TensorRT/tests/py/dynamo/distributed/test_native_nccl.py	2026-07-15 05:32:11.252442+00:00
+++ /home/runner/work/TensorRT/TensorRT/tests/py/dynamo/distributed/test_native_nccl.py	2026-07-15 05:32:39.841891+00:00
@@ -1365,10 +1365,11 @@

        This class is only reached when ENABLED_FEATURES.torch_tensorrt_runtime=False
        (C++ extension not installed), so setup_engine() naturally instantiates
        _TRTEngine without any patching needed.
        """
+
        class LinearModel(nn.Module):
            def forward(self, x: torch.Tensor) -> torch.Tensor:
                return x + 1

        model = LinearModel().cuda().eval()
@@ -2149,12 +2150,16 @@
    tests = [
        _multirank_all_reduce_correctness,
        _multirank_all_gather_correctness,
        _multirank_reduce_scatter_all_reduce_ops,
        _multirank_all_to_all_correctness,
-        lambda r, ws, dev: [_multirank_scatter_correctness(i, r, ws, dev) for i in range(ws)],
-        lambda r, ws, dev: [_multirank_gather_correctness(i, r, ws, dev) for i in range(ws)],
+        lambda r, ws, dev: [
+            _multirank_scatter_correctness(i, r, ws, dev) for i in range(ws)
+        ],
+        lambda r, ws, dev: [
+            _multirank_gather_correctness(i, r, ws, dev) for i in range(ws)
+        ],
        _multirank_distributed_mode_tp_model,
        _multirank_distributed_mode_subgroup,
        _multirank_cpp_runtime_bind_nccl,
        _multirank_distributed_mode_context_switch,
        _multirank_pg_migration,