[FX] Fix tests warnings

[daniil-lyakhov]

Changes

Fix Fx tests warnings

• UserWarning: Creating a tensor from a list of numpy.ndarrays is extremely slow.
• UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.detach().clone()

Reason for changes

To fix warnings

Related tickets

CVS-178300

Tests

[Copilot]

Pull request overview

Fixes two categories of PyTorch UserWarnings in FX-related tests by converting list-of-ndarrays to a single numpy array before tensor creation, and replacing torch.tensor(existing_tensor) with .detach().clone().

Changes:

• Wrap list of ndarrays with np.array(...) before torch.tensor(...) in statistics aggregator test fixture.
• Replace torch.tensor(tensor_input) with tensor.detach().clone() (with .to(dtype=...) where needed) in weights compression tests.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
tests/torch/fx/test_statistics_aggregator.py	Avoid slow tensor creation warning by converting list of ndarrays via np.array.
tests/torch/function_hook/quantization/test_weights_compression.py	Use .detach().clone() instead of torch.tensor(tensor) to avoid copy-construct warning.