Sudoku solver originally prepared for Kiwi.com coding challenge

Sudoku solver for 9x9 16x16 size.

Usage: main.py <input_file> <output_file>

Based on: https://techwithtim.net/tutorials/python-programming/sudoku-solver-backtracking/

Current speedup compared to original code (depending on size, and hardness of sudoku): ~50-67000x

Optimizations compared to original code from above:

1. Changed most for loops to list slicing & comprehension.
2. Changed for cycles to "in" for membership tests.
3. Moved size/box_size/etc constant calculations outside of for loops.
4. Added mask with possible valid values
5. Added board preprocessing based on valid values mask

New features:

1. Added ability to import csv
2. Added ability to solve both 9x9 & 16x16 sudoku puzzles
3. Added basic runtime timing for performance evaluation
4. Added logging for better performance evaluation
5. Added capability to save solutions to disk
6. Added concurrent batch processing capability

Todo:

1. Optimize further for speed? OR Write efficient parallel python code based on this paper: https://shawnjzlee.me/dl/efficient-parallel-sudoku.pdf
2. Test PyPy speedup?
3. Generalize for all possible sudoku sizes.
4. OCR
5. Multiplatform GUI with Kivy