JudgeBoard: Benchmarking and Enhancing Small Language Models for Reasoning Evaluation – The Association for the Advancement of Artificial Intelligence