Support emojis with modifiers and zero-width joiners

[Akron]

The tokenizer currently splits emoji sequences, either with modifiers or with ZWJs.
Examples are

• https://emojipedia.org/raised-fist-dark-skin-tone#technical (U+270A U+1F3FF)
• https://emojipedia.org/family-man-man-boy#technical (U+1F468 U+200D U+1F468 U+200D U+1F466).

In consequence this means for KorAP that queries need to treat these sequences as token sequences, like
[orth=✊][orth=🏿] and [orth=👨][orth=‍][orth=👦][orth=‍][orth=👦]. Which is especially problematic with ZWJ.

It's also therefore not possible to search for ✊🏿 or 👨‍👦‍👦.

This was reported by Louis Cotgrove.