Add a get device mesh(device type=None) method to ParallelismConfig that lazily builds and caches the device…

Add a get device mesh(device type=None) method to ParallelismConfig that lazily builds and caches the device…: a task in HF ML Tasksmith (Harbor dataset). Example: python cfg = ParallelismConfig(dp replicate size=8) assert cfg.device mesh is None mesh = cfg.get device mesh("cpu") builds and caches…