Judging the Judges: Evaluating Alignment and Vulnerabilities in LLMs-as-Judges