A brand new TUI and safe dataset caching

We've brought a lot of improvements to this template, making it a whole micro-framework for training and testing new models or datasets with PyTorch. This new release tag comes with the following additions:

• A restructuring of the project into something cleaner and easier to maintain.
• Integration of GitHub actions (static type checking, lining, formatting) and DeepSource for advanced code sanity checks.
• A brand new Text User Interface (TUI) based on Textual! This comes with a cleanup of the plotting and progress tracking for training/testing. It enables logging important messages into a scrollable view next to the training curves plot.
• A restructuring of the dataset base class into composable mixins. We're excited to deliver this feature!
  • You can now implement complex datasets with heavy preprocessing, all using multiprocessing with 0 headaches! We've all done it for you :)
  • Never mess up an experiment again! With the SafeDatasetCache mixin, we take care of the caching while making sure that your cache never gets stale! Changed a line in your dataset implementation? No worries, the smart cache will be flushed automatically :)

Beta release

First release

This micro-framework is ready to use, but further testing is needed. In the upcoming releases, unit testing and integration testing will be added.