amygdala lib: move_to_cpu=True to avoid bf16 SVD on CUDA

torch.svd doesn't support bf16 on CUDA; moving activations to CPU
first makes pca_aggregator work.

Co-Authored-By: Proof of Concept <poc@bcachefs.org>

training/amygdala_training/train_with_library.py

@@ -185,6 +185,7 @@ def main() -> None:
             aggregator=aggregator,
             batch_size=args.batch_size,
             show_progress=False,
+            move_to_cpu=True,
         )
         # sv.layer_activations is a dict {layer_idx: tensor[hidden]}
         for l_idx, layer in enumerate(target_layers):