reject non-BMP code points in the Character type converter

[digi-scrypt]

1. the Character converter parses a \uXXXX escape and returns Character.toChars(codePoint)[0], so a supplementary code point like \u1F600 silently becomes the lone high surrogate U+D83D.
2. that half surrogate is an invalid char and gets lost downstream (it round-trips to '?' through UTF-8).

Checked the code point is in the BMP before the cast and throw otherwise, so the value is rejected at conversion instead of quietly truncated. Have we considered that a char simply cannot hold an astral code point? Added a parameterized case for \u1F600 next to the existing \u0124 one.