One inference YAML per model; other accels commented under masking

Sync inference YAMLs from projects (init2 target_acceleration from val-4x)

Fix inference YAMLs: pin single R+ACS; document training schedule + how to override

Replace configs with minimal inference-only YAMLs (no training/validation)

Add E2E-ADS-Recon checkpoints (name.yaml + name.pt) for CMRxRecon / MIDL 2026