Added two base classes for the tabular models and allowed for bias and activations for the categorical embeddings

Against all good practices, this is a massive commit that adds an entire new module with Bayesian models (that are still not functional). Making them functional requires some work. Also the entire models module has been restrucured in preparation for better days to come

First commit towards v2. Re-organized the models module and added a few new functionalities for the models in there

Added docs. Re-design Additive Atttention accoring to their code. Add a few details on the docs and fix a typo

Added a proper implementation of transformer models when needed and re-organised the transformers module for clarity

First commit towards the perceiver. Changed modularity so that embeddings are consistent accross models. Adding the perceiver also forced me to change the structure of the TabTransformer and SAINT