CLI Generate.py ouput folder structure

When one goes to run the `jaxtrain.py` script using the CLI, one of the arguments is `--training-data-folder`, which is the path to the folder with the training data. At the moment, the generated data from `generate.py` is stored within a deep folder structure, for example: `./my_generated_data/data/training_data/lan/training_data_n_samples_20000_dt_0.001/ddm`

Ideally, it would be better for the metadata about the data generation from `generate.py` to be stored in another way, because when writing the CLI command for `jaxtrain` you have to point to the full path for the training data folder (above) rather than the much simpler `./my_generated_data`

One idea is to store the metadata about `n_samples`, `dt`, and the model type in the pickle file name instead.

Start by reading generate.py and jaxtrain.py, especially the --training-data-folder handling and the current generated-data layout. Trace how metadata such as n_samples, dt, and model type is written and discovered. Done means the CLI can use a simple generated-data path while retaining enough metadata for jaxtrain.py to locate the training data.

Written by the indexing model from the issue text.