ViTA: A Vision Transformer Inference Accelerator for Edge Applications