This PR adds required changes for XeGPU ops to support the SIMT distribution. 1. Adds verification logic for SIMT flavor for load_nd, store_nd, dpas, load_gather and store_scatter ops. 2. Adds test cases to cover the SIMT version of these ops along with their VC counter parts. --------- Co-authored-by: Artem Kroviakov <71938912+akroviakov@users.noreply.github.com>