Fix disable validation

[luigibonati]

Description

This PR fixes #83

Todos

I changed the DictModule as to accept a list with just one split. Furthermore, since by default the trainer expects a validation loader, it has to be disabled in the trainer options (see Lightning-AI/pytorch-lightning#12374)
I added these instructions in the tutorial on how to customize training

Status

• Ready to go

[codecov]

Codecov Report

Merging #88 (2a43254) into main (88b0189) will decrease coverage by 0.04%.
Report is 4 commits behind head on main.
The diff coverage is 83.33%.

[andrrizzi]

Looks good to me!

[luigibonati]

I did not merge this yet as I realized there is an issue with the logging, in that the MetricsCallback that I wrote is called only at the end of validation, so it does not log the training values. needs to fix that before merging

https://github.com/luigibonati/mlcolvar/blob/e356f2436568ca0361ec227cef3c33e332959bf0/mlcolvar/utils/trainer.py#L29C11-L29C11

[luigibonati]

I fixed the issue by changing the moment in which the callback is called (at the end of the training and not validation), merging it now