[CI] [tensornet] Attempt to set target causes fatal error in some OS environments

• The tensornet and tensornet simulators may not be available in the CI environment. A Python test which attempts to check the availability as shown below, causes the Validate wheel and Validate installer stages to crash.

(Added following in python/tests/kernel/test_sample_kernel.py)

@pytest.mark.parametrize("target", [
    'density-matrix-cpu', 'nvidia', 'nvidia-fp64', 'nvidia-mqpu',
    'nvidia-mqpu-fp64', 'qpp-cpu', 'tensornet', 'tensornet-mps'
])
def test_simulators(target):
    """Test simulation on all available simulation targets."""

    def can_set_target(name):
        target_installed = True
        try:
            cudaq.set_target(name)
        except RuntimeError:
            target_installed = False
        return target_installed

    if can_set_target(target):
        test_simple_sampling_qpe()
        cudaq.reset_target()
    else:
        pytest.skip("target not available")

    cudaq.reset_target()

Runs: locally in dev-image.
Validate wheel stage

• runs successfully on Fedora (arm64 and amd64), Debian (amd64)
• crashes on Ubuntu (amd64), RedHat (amd64), OpenSuse (amd64)

Validate installer stage

• only works on Debian (amd64)
• crashes on Ubuntu (arm64 and amd64), RedHat (arm64 and amd64), OpenSuse (amd64), Fedora (arm64 and amd64)

PR to showcase this behavior: https://github.com/NVIDIA/cuda-quantum/pull/1923/

CI: https://github.com/NVIDIA/cuda-quantum/actions/runs/9914417384?pr=1923

Start with python/tests/kernel/test_sample_kernel.py and the Validate wheel and Validate installer CI stages. Reproduce the simulator availability check across the listed operating systems, then use the PR and CI run linked in the issue to compare the failing behavior with the successful Debian and Fedora runs. Done means unavailable targets no longer crash either validation stage.