Shifting the retinal foundation models paradigm from slices to volumes for optical coherence tomography | Synapse