Make code and tests flexible to the use of a pre-processed Montgomery dataset...

Make code and tests flexible to the use of a pre-processed Montgomery dataset to speed-up testing where possible (closes #74 (closed)).

This MR also:

• Removes unused test data (closes #75 (closed))
• DRY on the make_split() function available in many data module instances
• Implements a method to easily re-write reference histograms for testing
• Add a script to help pre-processing the Montgomery dataset using the same code-base as mednet
• Increases the default line length to 88 (instead of 80 to match other packages in the software group)
• Remove unused _make_balanced_random_sampler() from the base datamodule module
• Remove the setting of any custom sampler objects on pytorch data loaders
• Remove the unused CSVDatabaseSplit from the code
• Mark slows tests (that that take more than 1 minute to execute)
• Provide pixi task test-fast to run test suit avoiding tests marked as slow; Also provide a test-slow to only run slow tests (closes #73 (closed))