String grapheme splitting support

We are missing a string grapheme splitting algorithm, causing string functions such as `length` (which originally counts graphemes), `split` (which splits into graphemes when the separator is empty), as well as `pop_grapheme` and `to_graphemes`, to be inconsistent with other targets.

We'd have to investigate the best way forward here. It's likely we'd have to make a small Rust or Python (or Gleam!) program to process Unicode data and generate relevant tables.