[Bug] TestTorchTensorRTModule::test_get_layer_info AssertionError on H100 with CUDA 13.x

Bug Description

TestTorchTensorRTModule::test_get_layer_info fails with AssertionError: False is not true on H100 GPUs across multiple CUDA 13.x versions. The failure is consistent across r13.2.0 and r13.1.1.

Environment

• GPU: H100
• Arch: x86_64
• CUDA: 13.2.0 / 13.1.1
• OS: Ubuntu 24.04
• cuDNN: 8.9.7.29
• TensorRT: 10.16.0.59
• TensorRT (Myelin): 2.17.78+7
• CASK: 5.16.17+1
• Python: 3.12
• Package: qa_tar_py3.12

Failing Test

FAILED api/test_classes.py::TestTorchTensorRTModule::test_get_layer_info - AssertionError: False is not true

Reproducible Configurations

Steps to Reproduce

1. Run on an H100 with CUDA 13.x and the environment listed above
2. Execute: pytest api/test_classes.py::TestTorchTensorRTModule::test_get_layer_info

Expected Behavior

test_get_layer_info should return valid layer information and the assertion should pass.

Additional Context

The test suite overall is healthy (46 passed, 1 skipped), with only this single test failing. The error message (False is not true) suggests get_layer_info() may be returning an empty or falsy result on CUDA 13.x, possibly due to an API change or missing support in the newer CUDA/TensorRT stack.

Start by running pytest api/test_classes.py::TestTorchTensorRTModule::test_get_layer_info on an H100 with the listed CUDA 13.x environment. Inspect the failing assertion and get_layer_info() behavior in api/test_classes.py, then verify that the test returns valid layer information and the full relevant test set passes.