fix: Update dependency fft-conv-pytorch to torch-fftconv

[vmiller987]

Hello all!

With nnUNet, we currently receive the following warning using PyTorch 2.9:

/home/vmiller/Work/caml/.venv/lib/python3.13/site-packages/fft_conv_pytorch/fft_conv.py:139: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:345.)
  output = output[crop_slices].contiguous()

The source of this is from fft-conv-pytorch which is a dependency of batchgeneratorsv2. Unfortunately, it does not look like fft-conv-pytorch is being maintained. This PR has existed since July: fkodom/fft-conv-pytorch#26

The offending code:
fft-conv-pytorch/fft_conv_pytorch/fft_conv.py lines 134-139`

    # Remove extra padded values
    crop_slices = [slice(None), slice(None)] + [
        slice(0, (signal_size[i] - kernel.size(i) + 1), stride_[i - 2])
        for i in range(2, signal.ndim)
    ]
    output = output[crop_slices].contiguous()

With the help of a coworker, we found an active fork with PyPi releases where someone did fix the issue:

• https://pypi.org/project/torch-fftconv/
• https://github.com/yoyolicoris/fft-conv-pytorch/tree/f80c6973cb2155c4844e81a784e50b2777e8ac2b

The fix:
fft-conv-pytorch/torch_fftconv/functional.py lines 242-245

    # Remove extra padded values
    index = (slice(None),) * 2 + tuple(slice(p, o + p)
                                       for p, o in zip(padding, output_size))
    output = output[index].contiguous()

I have updated and tested the two locations where this library is used in batchgeneratorsv2 and updated the pyproject.toml to replace the out of date dependency.