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

Cleaned and refactored the code for the attentive mlps. Adjusted unit tests. Docs build. Ready to move to the Bayesian mlp

Pmulinka/auc - a small fix related to torchmetrics

unit tested tabmlp, tabresnet, tabnet. Left are the transformers, widedeep and new unit tests for the attentive mlp

Pmulinka/losses - all additional changes by @jrzaurin are ok

Fix some minor type and style issues. Adjusted the ZILN Loss so it works with former versions of pytorch and rename quantile regression multilabel label as qregression

Embedding, MC and draft requets

Fixed #60 and #61. Placing dropout in the right position and applying FullEmbeddingDropout only in training

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

need consultation, I am not sure if I got it correctly