Multi-Head Explainer: A General Framework to Improve Explainability in CNNs and Transformers