Mini refactor

[gordicaleksa]

Mini refactor:

• Add informative error messages to each assert statement
• Cluster imports into python native & pip installed ones, sort alphabetically
• Make init argument order the same as self member assignment in NgramModel for consistency sake
• Add tqdm for tracking progress -> con: new dependency, pro: much nicer to understand the run progress
• Add caching of best hyperparams (mini optimization)
• Add get_prob_distribution to the api
• Use f-strings to print
• Use map for repetitive work

Additionally:

• Add environment.yaml so we can use a single python command to install all dependencies into a conda env. Otherwise users will additionally have to install matplotlib and ipykernel if they run the jupyter notebook script. This additionally allows us to nicely track the versioning information in the future.
• Add .gitignore