Properly handle unicode when splitting words

[sidharthv96]

1. Unicode/Emojis. Things like 🤦🏼‍♂️ are a combination of multiple characters. So we can't really use simple slice for the strings as meanings of things might get lost. Non Emoji Unicode combination characters will also have this problem.

2. Performance. For every single word, we're injecting an SVG into the DOM, removing it, again and again, character by character. So the default example mindmap would mean around 350 additions and 350 deletions into the DOM on a single render. There are programs which render on type, so 700+ DOM operations when the user types a single character. I will tweak the algorithm to only split if necessary.

Originally posted by @sidharthv96 in #4416 (comment)