Using `fold=all` does not work with the extension

[MathijsdeBoer]

One of the more "hidden" features of nnUNet is training with -f all, which uses all the available data for training, and validation. Using all in the fold box causes a failure to predict, as the plugin assumes that the fold is always a number.

Current workaround is to rename the all fold in the file tree to some number and rerunning the prediction.

[Thibault-Pelletier]

@MathijsdeBoer thank you for reporting this issue.
I will try to add this feature in coming versions of the extension.