Move etymology lookup to wiktionary

[dgw]

This spelling error prevents etymology lines from matching, because the code sets the mode to etymology but then later looks for a mode equal to etmyology. Fixed.

[dgw]

This actually accomplishes nothing, because the returned etymology from wikt() is never used. Perhaps this PR will morph into a pseudo-fix for #1248 (moving the .ety command into wiktionary.py)?

[dgw]

Built a proof-of-concept replacement .ety command on this branch, since the original PR was useless.

The wiktionary module only parses definitions for the first sense of a word, and the new .ety command implementation only looks for the first etymology (if it exists) to match that. A future rewrite (either in this PR or another, not yet decided) will add handling of multiple senses to both commands.

The Coveralls check failure for -0.9% test coverage tells me that my rewrite should also add example tests to this module. 12% coverage is downright awful. (The large drop must be from removing etymology.py, which was technically "30% covered" just from defining functions & initializing a ton of global variables.)

(Future implementation notes)

Good words for testing this future rewrite include "sake" and "forte". As of writing this comment, both include two separate senses with independent etymologies and their own sets of definitions. (Edit: TIL you can't use Markdown inside <details> blocks… Thanks, GitHub.)

[dgw]

6.6.0 will have #1432 to fix #1248, and this rewrite will maybe happen later—or not.

[dgw]

Shelving this until 7.1, unless etymology breaks again between now and 7.0's release.

[dgw]

I have no idea why I left this as WIP. It clearly works:

<dgw> ,ety bailiwick
<SopelTest> bailiwick: From bailie (“bailiff”) and wick (“dwelling”), from Old
            English wīc.

The main difference is obviously how detailed the etymologies are, but I think Wiktionary is a better bet over the long term. WT's editors will continue to flesh it out over time.

I have a conflicts-resolved merge already done locally (it's not that bad) for testing, so don't worry about GitHub showing that this isn't mergeable @ reviewers.