Always use most accurate way of fuzzy matching, greatly improve performance of fuzzy matching

[aaronjwood]

I found the Bash implementation of calculating the Levenshtein distance to take between ~2-5 seconds depending on the input, using the Python module (important to note that it uses Cython) is MUCH faster:

time python -c 'from pyxdameraulevenshtein import damerau_levenshtein_distance; print(damerau_levenshtein_distance("Very Long Album Title: The Hits Part 2 Vol. 2", "Some Other Album Part One: Vol 1."))'
32

real    0m0.009s
user    0m0.009s
sys     0m0.000s

time python -c 'from pyxdameraulevenshtein import damerau_levenshtein_distance; print(damerau_levenshtein_distance("Similar Album Title: Part One Vol 1.", "Similar Album Title: Part Two Vol 2."))'
4

real    0m0.015s
user    0m0.012s
sys     0m0.004s

Because of this I felt we could remove the rough-but-fast-and-early way of detecting if there's a match.