Optimised Reasoning for {$\mathcal SHIQ$}