Include MAIA tasks into Evalita tasks

Add standard deviation of accuracy values across prompts

Added results for Fastweb, HuggingFaceH4, and Gemma-2-27B

Minor changes

Update the models result

Remove the whole dataset

Initial commit