Fix: handle CUDA tensors in df_features() by moving audio to CPU before NumPy conversion

[F1xxs]

While working on a personal project using DeepFilterNet with an external model operating on CUDA tensors, I encountered the following error:

spec = df.analysis(audio.numpy())  # [C, Tf] -> [C, Tf, F]
TypeError: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.
^C

To resolve this, I applied .detach() and .cpu() on the audio tensor to remove it from autograd graph and move to the CPU.

spec = df.analysis(audio.detach().cpu().numpy())