Structure data folder

[gcroci2]

We started symlinking data on Snellius with folders on 3D-Vac repo, but it was a bit messy and at the end we didn't finish the job.

For now, we're going to keep the exact data structure that we have on Snellius, soft linking data. See issue #47

In the future, we're probably going to restructure it in something like this:

├── data
│   │
│   ├── hdf5
│   │   ├── cnn         <- hdf5 files generated with old version of deeprank, for cnns training
│   │   └── gnn         <- hdf5 files generated with new version of deeprank, for gnns training
│   │
│   ├── 3d_models       <- aligned 3d models, output of pandora
│   │
│   ├── external
│   │   ├── processed   <- csv files ready for the modelling
│   │   └── unprocessed <- BA and EL text data (csv)
│   │
│   └── pssm
│       ├── mapped
│       ├── unmapped
│       └── blast_dbs   <- blast databases and MSAs to generate them

[gcroci2]

This is outdated. In case in which we'll change the snellius data structure again we'll open a new issue.