Process extremely large chess PGN files

Beware, this project is very rough at the moment!

Working, sort of:

• Split arbitrarily large PGN files into smaller files based on moves played, player rating, time control, etc.
• Create a nested directory structure containing the sub-PGNs, with the directory structure representing the tree of moves played.
• Find and merge transposing positions.
• Create Sankey flow diagrams of chess openings.

Known issues:

• The Sankey diagrams don't handle tranpositions well.
• The code is generally finicky.

Goals:

• Cleanup ugly and/or inefficient code.
• Improve workflow/usability.
• Add engine analysis to positions.
• Integrate engine analysis with data from the PGN such as win probability to generate opening repertoires.

Current Workflow:

• (Currently optional) Use sanitize.py to "clean" a large PGN database (removes annotations, commentary, engine analysis, etc to greatly reduce filesize and make processing easier.)
• If a randomish sample from a large PGN db is desired, creat the sample with sample.py
• If one opening in particular is of interest, create a PGN of games with only that opening using get_opening.py.
• Use split.py to iteratively split a PGN out by opening moves.
• If needed to free up storage space, remove the sub-PGNs with del_pgns.py
• Create charts with sankey.py