use plover_stroke

[benoit-pierre]

Summary of changes

The next version of plover_stroke (1.0.0) has been rewritten to use a CPython extension, making it possible to use it for the steno module implementation without negatively impacting the performance (particularly loading times).

Pros:

• one consistent implementation: same behavior (including normalization) for steno.Stroke, normalize_steno, and plover_build_utils.testing.steno_to_stroke
• better steno normalization (see tests), including detection of invalid strokes (used by the "add translation" dialog and dictionary editor to prevent adding invalid mappings)
• new steno_to_sort_key for sorting on steno order (used by the dictionary editor)
• slightly faster loading times (~1.5 faster at loading main.json)

Cons:

• plover_stroke support a maximum of 63 keys
• using a CPython extension means no universal wheel for plover_stroke (but binary wheels for common platforms / Python versions are provided)
• sorting on steno order is slower than the standard string sort

TODO:

• use steno_to_sort_key for sorting on JSON dictionary save
• use the framework added in rework dictionaries widget #1332 do add a test for gui_qt.steno_validator

Pull Request Checklist

• Changes have tests
• News fragment added in news.d. See documentation for details

[benoit-pierre]

I think this take care of #672.

[panathea]

Looks good to me!