refactor(multibyte): replace generated unicode tables with utf8proc

This commit intentionally aims at preserving existing behavior as much
as possible while replacing our build step to convert unicode data
files into binary tables, which corresponding lookups in utf8proc.
Actual improvements in behavior will be a followup.

The only change in behavior is that 'emoji' option will turn some
more codepoints into double with. Nvim used the "Emoji" and
"Emoji_Presentation" properties to define emojis, while utf8proc
only exposes the Extended_Pictographic property from the emoji table.
This is a superset of the previous emoji properties. As only
codepoints above 0x1f000 are affected by the 'emoji' option, this means
that the following chars are now treated as double-width, instead of
single-width like in previous nvim versions:

🀀 🀁 🀂 🀃 🀅 🀆 🀇 🀈 🀉 🀊 🀋 🀌 🀍 🀎 🀏 🀐 🀑 🀒 🀓 🀔 🀕 🀖 🀗 🀘 🀙 🀚 🀛 🀜 🀝 🀞 🀟 🀠 🀡 🀢 🀣 🀤
🀥 🀦 🀧 🀨 🀩 🀪 🀫 🀰 🀱 🀲 🀳 🀴 🀵 🀶 🀷 🀸 🀹 🀺 🀻 🀼 🀽 🀾 🀿 🁀 🁁 🁂 🁃 🁄 🁅 🁆 🁇 🁈 🁉 🁊 🁋 🁌
🁍 🁎 🁏 🁐 🁑 🁒 🁓 🁔 🁕 🁖 🁗 🁘 🁙 🁚 🁛 🁜 🁝 🁞 🁟 🁠 🁡 🁢 🁣 🁤 🁥 🁦 🁧 🁨 🁩 🁪 🁫 🁬 🁭 🁮 🁯 🁰
🁱 🁲 🁳 🁴 🁵 🁶 🁷 🁸 🁹 🁺 🁻 🁼 🁽 🁾 🁿 🂀 🂁 🂂 🂃 🂄 🂅 🂆 🂇 🂈 🂉 🂊 🂋 🂌 🂍 🂎 🂏 🂐 🂑 🂒 🂓 🂠
🂡 🂢 🂣 🂤 🂥 🂦 🂧 🂨 🂩 🂪 🂫 🂬 🂭 🂮 🂱 🂲 🂳 🂴 🂵 🂶 🂷 🂸 🂹 🂺 🂻 🂼 🂽 🂾 🂿 🃁 🃂 🃃 🃄 🃅 🃆 🃇
🃈 🃉 🃊 🃋 🃌 🃍 🃎 🃑 🃒 🃓 🃔 🃕 🃖 🃗 🃘 🃙 🃚 🃛 🃜 🃝 🃞 🃟 🃠 🃡 🃢 🃣 🃤 🃥 🃦 🃧 🃨 🃩 🃪 🃫 🃬 🃭
🃮 🃯 🃰 🃱 🃲 🃳 🃴 🃵 🄍 🄎 🄏 🄯 🅬 🅭 🅮 🅯 🆭 🌢 🌣 🎔 🎕 🎘 🎜 🎝 🏱 🏲 🏶 📾 🕆 🕇 🕈 🕏 🕨 🕩 🕪 🕫
🕬 🕭 🕮 🕱 🕲 🕻 🕼 🕽 🕾 🕿 🖀 🖁 🖂 🖃 🖄 🖅 🖆 🖈 🖉 🖎 🖏 🖑 🖒 🖓 🖔 🖗 🖘 🖙 🖚 🖛 🖜 🖝 🖞 🖟 🖠 🖡
🖢 🖣 🖦 🖧 🖩 🖪 🖫 🖬 🖭 🖮 🖯 🖰 🖳 🖴 🖵 🖶 🖷 🖸 🖹 🖺 🖻 🖽 🖾 🖿 🗀 🗁 🗅 🗆 🗇 🗈 🗉 🗊 🗋 🗌 🗍 🗎
🗏 🗐 🗔 🗕 🗖 🗗 🗘 🗙 🗚 🗛 🗟 🗠 🗢 🗤 🗥 🗦 🗧 🗩 🗪 🗫 🗬 🗭 🗮 🗰 🗱 🗲 🗴 🗵 🗶 🗷 🗸 🗹 🛆 🛇 🛈 🛉
🛊 🛓 🛔 🛦 🛧 🛨 🛪 🛱 🛲 🝴 🝵 🝶 🝻 🝼 🝽 🝾 🝿 🟕 🟖 🟗 🟘 🟙 🢰 🢱 🨀 🨁 🨂 🨃 🨄 🨅 🨆 🨇 🨈 🨉 🨊 🨋
🨌 🨍 🨎 🨏 🨐 🨑 🨒 🨓 🨔 🨕 🨖 🨗 🨘 🨙 🨚 🨛 🨜 🨝 🨞 🨟 🨠 🨡 🨢 🨣 🨤 🨥 🨦 🨧 🨨 🨩 🨪 🨫 🨬 🨭 🨮 🨯
🨰 🨱 🨲 🨳 🨴 🨵 🨶 🨷 🨸 🨹 🨺 🨻 🨼 🨽 🨾 🨿 🩀 🩁 🩂 🩃 🩄 🩅 🩆 🩇 🩈 🩉 🩊 🩋 🩌 🩍 🩎 🩏 🩐 🩑 🩒 🩓
🩠 🩡 🩢 🩣 🩤 🩥 🩦 🩧 🩨 🩩 🩪 🩫 🩬 🩭

0 files changed, 0 insertions, 0 deletions