Assessing the Feasibility of Lightweight Whisper Models for Low-Resource Urdu Transcription