Tag fuzzy matching and aliasing

[sco1]

As discussed briefly in #118 and at least a few times on the server in the recent past, I believe tags could stand to gain from fuzzy matching and possibly from aliasing as well.

In short:

• Fuzzy Matching

  • Allows for solving of the plurals problem, like codeblock vs codeblocks and f-string vs f-strings
  • Potentially solves things like fstring vs f-string
  • Embed should indicate to the user that a fuzzy match was used (e.g. "Did you mean ... ?")

• Aliasing (See Tag aliases #187)

  • Requires adding a field to the backend tag model
  • Allows for things that fuzzy matching may not realistically be able to hit, like ytdl vs youtube-dl

[jb3]

As mentioned in the existing issue for tag aliases this should be done after the Django rewrite, and potentially after we split up the bot.

[jb3]

I believe that the previous idea was supporting plurals of all tags automatically and then having aliases in place to handle cases that weren't just about the plurals. Fuzzy matching could help but I don't know if it essential.

[jchristgit]

With the site migration to PostgreSQL, I can look into using pg_trgm which supports a very nice way to search for similar items using the % operator ("x is similar to y"):

postgres=# SELECT 'f-strings' % 'f-string';
 ?column? 
----------
 t
(1 row)

postgres=# SELECT 'ask' % 'f-strings';
 ?column? 
----------
 f
(1 row)

The threshold is configurable, but this should be able to return accurate results.