Reorganize DataModule and Dataset Configuration Structure

[Oufattole]

Current Structure

Code Organization

• The DataModule class in src/meds_torch/data/datamodule.py can contain either:
  • An instance of pytorch_dataset class
  • An instance of multiwindow_pytorch_dataset class
• Manual mapping is currently defined here

Configuration Files

Current config files combine settings for both DataModule and Dataset classes:

src/meds_torch/configs/data/
├── pytorch_dataset.yaml
└── random_windows_pytorch_dataset.yaml

Proposed Changes

New Configuration Structure

Separate dataset-specific configs from DataModule configs:

src/meds_torch/configs/data/
├── datamodule.yaml
└── dataset_type/
    ├── pytorch_dataset.yaml
    └── random_windows_pytorch_dataset.yaml

Benefits

The main reason for this is so that users can use their own custom dataset files and configs, and select them via hydra.

Implementation Tasks

• [ ] Create new directory structure
• [ ] Split existing YAML files into appropriate components
• [ ] Update code references to new config paths
• [ ] Verify all dataset types work with new structure