Change `TaggedTable._into_dataloader_with_classes` to use `to_numpy()` instead of `values`

[Marsmaennchen221]

Is your feature request related to a problem?

Currently, the method uses DataFrame.values to convert a pandas DataFrame into a PyTorch Tensor. It is recommended to use DataFrame.to_numpy() instead, as this does the same, but DataFrame.values could be deprecated soon. In the documentation from DataFrame.values it is also recommended to use DataFrame.to_numpy instead.

Desired solution

Change lines 902 and 911 from torch.Tensor(self.features._data.values), to torch.Tensor(self.features._data.to_numpy),

Possible alternatives (optional)

No response

Screenshots (optional)

No response

Additional Context (optional)

No response

[lars-reimann]

Polars, which we now use for our tabular containers, also has a direct DataFrame.to_torch.

[lars-reimann]

Has been done as part of #752.