Re-train spinal canal model and implement continual learning strategy

[jcohenadad]

Todo:

• Quality check the existing ground truth segmentations (some might be bad, lowering performance of the model). Could be selected based on the Dice after running inference with latest model.
• Throw a couple more datasets for training, incl. EPI (from Sandrine and Caroline)
• Select test set from heterogeneous database. Ideally public, so that SpinePS could also be evaluated
• Implement continual learning strategy to monitor model drift, as in https://pmc.ncbi.nlm.nih.gov/articles/PMC12828353/