Make docs better for new data

It's not documented well how to run a model with your own dataset, especially the part about the drug input. E.g., it doesn't state clearly how to generate the drug features and where the fingerprints should be located and how the file should look like (drug ids = column names). It also doesn't state clearly that you should adapt the meta/landmark_genes_reduced.csv, especially for data coming from a different species.

Start by locating the documentation for running a model with a custom dataset and review how drug features, fingerprints, and drug IDs are described. Clarify how to generate the features, where fingerprints belong and how their file should be structured, and explain when to adapt meta/landmark_genes_reduced.csv, including for data from another species.

Written by the indexing model from the issue text.