MicroScopiQ: Accelerating Foundational Models through Outlier-Aware Microscaling Quantization