Increase max tokens and temperature for richer clinical narratives

Switch to IdaFLab/ECG-Plot-Images Parquet dataset to prevent HTTP bottlenecks

Switch to edcci/GenECG public image dataset to avoid MIMIC-IV protected paths

Switch to massive PULSE-ECG real world dataset

Add robust fallback generation for mock ECG datasets

Add GitHub Actions workflow

Temp remove workflow

Initialize CardioAI training and deployment pipeline