General to-dos

[tomoleary]

• Add license and boilerplate to everything.
• Separate the data generation from the training pipeline

[joshuawchen]

Data generation and training are already separated:

Data generation: https://github.com/dinoSciML/dinox/blob/64ee6f22004eb93516cafa7271775625f36299d2/src/dinox/data_generating.py#L253C5-L253C43

Training:
Either: https://github.com/dinoSciML/dinox/blob/64ee6f22004eb93516cafa7271775625f36299d2/src/dinox/training_utils.py#L275C5-L275C42

or
https://github.com/dinoSciML/dinox/blob/64ee6f22004eb93516cafa7271775625f36299d2/src/dinox/training_utils.py#L335C5-L335C24

(depending on whether one wants to manually handle the loading of data)

[joshuawchen]

dinox uses the MIT license already.