make discord bot produce several markov results and select most relevant

[Trogluddite]

currently the /search command in the discord bot doesn't actually search anything. It gets a few markov chains, then formats them into a sort of paragraph.

To make this searchy, we want to do something like this:

1. generate several (10 or so) candidate 'paragraphs' from the MatrixMarkov instance
2. compare the results to some target string, using some heuristic.
   • The simplest is probably just searching for any instances of that particular n-gram
   • more complex is computing the Levenshtein distance (https://en.wikipedia.org/wiki/Edit_distance) -- you'd probably want to tokenized the markov-text into n-grams of the same length as the input text, then compute the similarity for each n-gram
   • more complex still, but I think out of scope for this project, would be a sort of inverse of the method we use to produce markov text -- a perception classifier. Something like this: https://www.linkedin.com/pulse/single-perceptron-based-text-classifier-applied-analysis-daboubi/.
3. based on that heuristic, return the top-ranked (or top-n) result(s) to the user.